The DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index; none of the 20 sources its claims cite is checked for support yet.
The need for uncertainty quantification in machine-assisted medical decision making
Open accessOverview
AI snapshot
Written from the paper’s full text.
TL;DR
This perspective argues that machine-assisted medical decision making, especially deep learning, requires a principled and formal discipline of uncertainty quantification (UQ) comparable to that used in nuclear stockpile stewardship and risk management. It maps where uncertainty enters the DL pipeline, identifies four overlapping challenge areas, and sketches research directions for building UQ for DL.
Key findings
- The paper contends that predictions without UQ "are neither predictions nor actionable," and that automated clinical systems not analyzable via known causal connections will require formal UQ to be adopted, as happened in nuclear security.
- It identifies four overlapping challenge groups for UQ of data-driven approaches: (1) absence of theory (no "hard laws" in medicine), (2) absence of causal models (DL exploits correlations without causal links), (3) sensitivity to imperfect data (missing elements and errors with patterned confounds), and (4) computational expense of recomputation for uncertainty bounds.
- It proposes that uncertainty must be quantified across at least four pipeline steps: collection/selection of training data and its coverage of the medical phenomenon, accuracy and completeness of labelling, selection and understanding of the DL model and its performance bounds, and model performance against operational (clinical inference) data.
- It highlights three research directions: quantifying and limiting overfitting (noting classical information criteria fail given DL's parameter counts and capacity to memorize random noise), understanding DL internals, and training DL to provide its own uncertainty estimates—suggesting GANs for detecting out-of-sample cases as "particularly fruitful."
- Cited (not original) performance claims: DL transfer-learning applications reported performance comparable to human experts or better, and DL function estimators reported performance better than state-of-the-art classical predictive models.
Methods
- Design: a conceptual/perspective article (no empirical study, dataset, or experiment conducted by the authors).
- Content: narrative discussion of the role of DL in clinical decision making, the proposed place of UQ, anticipated challenges, and foreseen research directions, illustrated by a figure of uncertainty sources in the DL pipeline.
- Analysis: argumentative synthesis drawing analogies to UQ practice in nuclear security and risk management; no statistical analysis is performed.
Limitations
- Stated by the paper: the domains where DL is applied lack underlying mechanistic theory to ground uncertainties; DL lacks causal models, limiting conclusions; real-world data is imperfect with patterned errors that can confound prediction; and recomputation for uncertainty bounds may currently be prohibitively expensive.
- Evident in the text: the article is a position/perspective piece with no empirical validation, no concrete UQ framework or method proposed, and no quantitative results of its own; the authors explicitly note that "it is possible that the entire new field of UQ for DL might need to be developed" and that a review of ongoing research is beyond its scope.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
29 Own interpretation · 17 No basis given · 10 Not checked yet
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“we are ultimately operating without the fundamental, underlying mathematical model which we could otherwise use to ground our uncertainties and to bound any assumptions we make.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4 Anticipated Challenges
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“Advancements in UQ now provide measures of confidence necessary to inform national or international security decisions.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
What the sources say
- Not checked yetcites [1] Oberkampf, W. L. & Roy, C. J. Verfication and Validation in Scientific Computing (Cambridge Univ. Press, Cambridge, 2010). source
- Not checked yetcites [2] National Research Council Evaluation of Quantification of Margins and Uncertainties: Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile (National Academies Press, Washington DC, 2009).
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The absence of a causal connection, however, means garnering limited conclusions from DL models”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: unclear — It is unclear whether it could be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“most partial theories in medicine still explain an extremely small fraction of the observed phenomena and variations 3,4 .”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2 Machine-assisted Clinical Decision Making and Research
What the sources say
- Not checked yetcites [3] Zuk, O., Hechter, E., Sunyaev, S. R. & Lander, E. S. The mystery of missing heritability: Genetic interactions create phantom heritability. Proc. Natl Acad. Sci. USA 109, 1193–1198 (2012). source
- Not checked yetcites [4] Choi, J. D. & Lee, J.-S. Interplay between epigenetics and genetics in cancer. Genomics Inform. 11, 164–173 (2013). source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“The applications of DL techniques based on transfer learning have reported performance comparable to that of the human experts 5 , , 6 or better 7 ., 8”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2.1 The Current Roles and Applications of AI
What the sources say
- Not checked yetcites [5] Kermany, D. S. et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell 172, 1122–1131 (2018). source
- Not checked yetcites [6] Esteva, A. et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature 542, 115–118 (2017). source
- Not checked yetcites [7] Mar, V. & Soyer, H. Artificial intelligence for melanoma diagnosis: How can we deliver on the promise? Ann. Oncol. 29, 1625–1628 (2018). source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“uncertainty in generalization depends on the density of training points in an appropriately defined neighborhood of the prediction target.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.3 Training DL to provide its own uncertainty estimates
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“A particularly fruitful approach seems to be the use of Generative Adversarial Networks (GANs) for detecting out-of-sample cases.[24][25][26]”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.3 Training DL to provide its own uncertainty estimates
What the sources say
- Not checked yetcites [24] Brahma, P. P., Huang, Q. & Wu, D. O. Structured memory based deep model to detect as well as characterize novel inputs. Preprint at http://arxiv.org/abs/1801.09859 (2018).
- Not checked yetcites [25] Yu, Y., Qu, W., Li, N. & Guo, Z. Open-category classification by adversarial sample generation. Preprint at http://arxiv.org/abs/1705.08722 (2017). source
- Not checked yetcites [26] Ge, Z., Demyanov, S., Chen, Z. & Garnavi, R. Generative openmax for multi-class open set classification. Preprint at http://arxiv.org/abs/1707.07418 (2017). source
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“the capacity of DL networks to memorize random noise 18”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.1 Quantifying and limiting overfitting
What the sources say
- Not checked yetcites [18] Zhang, C., Bengio, S., Hardt, M., Recht, B. & Vinyals, O. Understanding deep learning requires rethinking generalization. Preprint at http://arxiv.org/abs/1611.03530 (2016).
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Probably not a substantial claim. 44
- It may state two separate claims as one. 16
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“computing capacity in support of DL is growing exponentially”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“UQ today underpins many decision processes in nuclear security, our risk management and associated investments, which can be at the scale of billions of dollars.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“we propose that all of these steps (illustrated in figure 1) would need to be quantified in order to arrive at even crude, overall measures of the uncertainty of the DL-based decision model.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 3 The Role for UQ in DL
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“as DL-structured neural networks are particularly suitable for recognition of visually manifested conditions such as changes in tissue, lesions and growth, etc.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2.1 The Current Roles and Applications of AI
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Real world data is usually imperfect-typically containing missing elements and errors-and these imperfections have patterns that can confound prediction.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“the domains where the DL is usually applied, such as medicine, do not have such”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4 Anticipated Challenges
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We see at least four overlapping groups of challenges associated with the uncertainty quantification of the data-driven approaches such as DL:”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4 Anticipated Challenges
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“To achieve this confidence, UQ extends the traditional discipline of statistical error analysis to also capture uncertainties due to possibly incomplete, inaccurate, and contradictory input data, missing and undetected mechanisms and dependencies, expert judgment, and variations between reasonable model forms and modeling strategies.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“use of scientific insights and attempts at a cohesive framework incorporating the major clinical predictors are likely to be increasingly useful as predictive models are able to efficiently summarize more complex correlations in the data.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2 Machine-assisted Clinical Decision Making and Research
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Predictions without UQ are neither predictions nor actionable.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Although AI-based research has long played an important role in medicine, it has, nevertheless, been one of the more challenging areas for AI to see an effective adoption.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2 Machine-assisted Clinical Decision Making and Research
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
- Rests on an assumption: The reluctance to delegate to machine intelligence a function of life-critical and patient-safety-related decision making is a primary reason for the slow adoption of AI in medicine.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“In most of the presented cases, the applications of AI are based on the DL neural networks, trained on a very large number of labeled data sets, and their learning tuned with the large number of hyperparameters.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2.1 The Current Roles and Applications of AI
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“A need to systematize a similar situation is what actually led us to develop the formal approach to UQ in US national security sciences.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“medical AI, especially in its modern data-rich Deep Learning guise, needs to develop a principled and formal Uncertainty Quantification (UQ) discipline, just as we have seen in fields such as nuclear stockpile stewardship or risk management.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
- Rests on an assumption: The reluctance to delegate to machine intelligence a function of life-critical and patient-safety-related decision making is a primary reason for the slow adoption of AI in medicine.
None of the sources cited for this claim have been checked yet.
In the paper
“A notable example is the US support of a nuclear test moratorium since 1992, whereby we annually provide detailed measures of confidence in the safety, security, and performance of the nuclear stockpile--guaranteed through virtual testing 1,2 .”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
What the sources say
- Not checked yetcites [1] Oberkampf, W. L. & Roy, C. J. Verfication and Validation in Scientific Computing (Cambridge Univ. Press, Cambridge, 2010). source
- Not checked yetcites [2] National Research Council Evaluation of Quantification of Margins and Uncertainties: Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile (National Academies Press, Washington DC, 2009).
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The data-rich world of AI-based learning, and the frequent absence of a well-understood underlying theory poses its own unique challenges to straightforward adoption of UQ.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Obviously, there are many steps in such scenarios where there are uncertainties that would need to be quantified.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 3 The Role for UQ in DL
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“ad hoc solutions, such as sensitivity analysis and study of model variability, have sometimes been employed to mitigate some the problems we outline.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“once a low-dimensional representation is arrived at describing certain correlations (e.g., the difference between cancerous and matched normal cells), it can raise hypotheses that can be tested.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“These are then”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2.1 The Current Roles and Applications of AI
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“This may not seem like a limitation since prediction does not need causal relation.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Understanding prediction system structure and defensibly quantifying uncertainty is possible, and, if done, can significantly benefit both research and practical applications of AI in this critical domain.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Automated systems adoption, especially systems not analyzable in terms of known causal connections, will require principled and formal UQ to play a transformative role, just as we have seen in the nuclear security domain.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2 Machine-assisted Clinical Decision Making and Research
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
In the paper
“techniques are being developed 17 to approximate some of the UQ-relevant calculations.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
What the sources say
- Not checked yetcites [17] Gal, Y. & Ghahramani, Z. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. Preprint at https://arxiv.org/abs/1506.02142 (2015).
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“But numerical results, without a measure of their veracity do not provide the trust needed to inform decisions.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“The training of the DL models is computationally expensive, and any further recomputation and re-evaluation of the models, aimed, for example, at the calculation of uncertainty bounds might currently be prohibitively expensive.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These challenges, while not trivial, also present significant new research opportunities for the development of new theoretical approaches, and for the practical applications of UQ in the area of machine-assisted medical decision making.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“furthermore, it is imperative to understand how the training data must be similar to prediction data.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Specific UQ methods, therefore, need to be developed to quantify the sensitivity of models to imperfect data.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“DL is essentially exploiting correlations in the data, without paying attention to any causal link.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2. Absence of causal models:
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“To develop consequential decision-support from”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > Abstract
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Unlike statistical approaches where mathematical models are used to explain variations observed in data, and to propose the margin of errors on inferences, with these recent applications, different learning architectures are combined with a large number of DL network parameters to form universal approximators.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 2.1 The Current Roles and Applications of AI
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Data-driven methods are emerging as the foundations of evidence-based decision making, and the future of data-driven scientific discovery.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 6 Summary
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“Some approaches, such as attempts to empirically learn generalizable patterns with insertion of random noise 19,20 , or the use of cross-validation to determine the progression of generalizable learning, move us forward in this problem space, while still carrying the problem of overfitting 21 .”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.1 Quantifying and limiting overfitting
What the sources say
- Not checked yetcites [19] Arpit, D. et al. A closer look at memorization in deep networks. Preprint at https://arxiv.org/abs/1706.05394 (2017).
- Not checked yetcites [20] Zhang, C., Vinyals, O., Munos, R. & Bengio, S. A study on overfitting in deep reinforcement learning. Preprint at http://arxiv.org/abs/1804.06893 (2018).
- Not checked yetcites [21] Cawley, G. C. & Talbot, N. L. C. On over-fitting in model selection and subsequent selection bias in performance evaluation. J. Mach. Learn. Res. 11, 2079–2107 (2010).
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The wider application of DL in the biomedical field now requires an extension of these methods to this emerging field.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“an effective way of addressing some of the mentioned UQ problems might be to re-shape the DL engine itself to provide an uncertainty estimate on its predictions.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.3 Training DL to provide its own uncertainty estimates
hedgedconditional- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Overfitting, or the problem of a model performing well on the training set, but generalizing poorly for unseen datasets, is one of the fundamental problems of all data-centric methods, and therefore DL.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.1 Quantifying and limiting overfitting
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The details of the medical applications and DL networks are, however, significantly different since the theoretical foundations are far less developed and there are deep psychological and sociological implications in delegating to machines decisions that affect the life or health of human beings.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 6 Summary
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“progress in understanding the structure of these predictive systems, merging model and data driven approaches with strongly defensible UQ, and the formalization of the UQ for DL discipline will be needed to make DL and other data-centric tools and methods practically useful.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 6 Summary
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These have been studied in other scientific disciplines and decision-making arena, and we can learn from those.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 6 Summary
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“UQ for DL will likely not be simply a set of tools or procedures to apply, but a more complex wrap-up of disparate methods that in total help bound the overall confidence in predictions.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 6 Summary
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In high-dimensional problems, typical to the medical setting (images, large numbers of phenotypes, etc.), every point, however, can be isolated in some other dimension, and a density of points makes sense only after irrelevant dimensions, are projected out”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.3 Training DL to provide its own uncertainty estimates
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
In the paper
“With DL, and the large number of model parameters involved as well as the capacity of DL networks to memorize random noise 18 , classical approaches do not work.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.1 Quantifying and limiting overfitting
What the sources say
- Not checked yetcites [18] Zhang, C., Bengio, S., Hardt, M., Recht, B. & Vinyals, O. Understanding deep learning requires rethinking generalization. Preprint at http://arxiv.org/abs/1611.03530 (2016).
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Advances in understanding of how DL works internally will allow for a more effective UQ of interpreting DL.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.2 Understanding DL
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“there is a unifying structure to this problem in its various guises of complex predictive correlations in large data sets and engineering black boxes.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 6 Summary
plain- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper may give no reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“there are ongoing studies that attempt to understand what DL does, 22 and how it learns 23”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.2 Understanding DL
What the sources say
- Not checked yetcites [22] Brahma, P. P., Wu, D. & She, Y. Why deep learning works: A manifold disentanglement perspective. IEEE Trans. Neural Netw. Learn. Sys. 27, 1997–2008 (2016). source
- Not checked yetcites [23] Raghu, M., Gilmer, J., Yosinski, J. & Sohl-Dickstein, J. SVCCA: Singular vector canonical correlation analysis for deep learning dynamics and interpretability. Preprint at https://arxiv.org/abs/1706.05806 (2017).
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the network itself can be used to study this uncertainty empirically and provide the uncertainty bounds.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 5.3 Training DL to provide its own uncertainty estimates
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In the end, it is possible that the entire new field of UQ for DL might need to be developed.”
§ The Need for Uncertainty Quantification in Machine-assisted Medical Decision Making > 4. Computational Expense:
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: fails — It may be too vague to check against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. It may be too vague to check against evidence. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper relies on this without support.
The reluctance to delegate to machine intelligence a function of life-critical and patient-safety-related decision making is a primary reason for the slow adoption of AI in medicine.
“There are many reasons for this, primarily the reluctance to delegate to machine intelligence a function of life-critical and patient-safety-related decision making.”
The authors state this as an assumption.
Fair sampling in the training data holds.
“assuming fair sampling in the training data”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Better decision-making: AI, ML, and DL can enable robots to make better decisions based on data analysis and pattern recognition, leading to improved performance and outcomes [52] .”
From Artificial intelligence, machine learning and deep learning in advanced robotics, a review (2023) · cites this paper as [52]
Experimental — accuracy not yet measuredIn the citing paper
“However, these models have black-box nature and lack underlying theoretical foundation behind their decisions [8]”
From Explainable, trustworthy, and ethical machine learning for healthcare: A survey (2022) · cites this paper as [8]
In the citing paper
“However, due to the unavailability of sound fundamental laws and models, the explanation of DL cannot be produced up-to the required scale.”
From Explainable, trustworthy, and ethical machine learning for healthcare: A survey (2022) · cites this paper as [8]
Citations
Cites
Dermatologist-level classification of skin cancer with deep neural networks
Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. M., Blau, H. M., & Thrun, S. (2017) · Nature · doi:10.1038/nature21056 · Cited by 14,362
Dropout as a bayesian approximation: Representing model uncertainty in deep learning
Gal, Y., & Ghahramani, Z. (2015) · arXiv · doi:10.48550/arxiv.1506.02142 · Cited by 6,471
Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning
Kermany, D. S., Goldbaum, M., Cai, W., Valentim, C. C. S., Liang, H., Baxter, S. L., McKeown, A., Yang, G., Wu, X., Yan, F., et al. (2018) · Cell · doi:10.1016/j.cell.2018.02.010 · Cited by 4,842
Understanding deep learning requires rethinking generalization
Zhang, C., Bengio, S., Hardt, M., Recht, B., & Vinyals, O. (2016) · arXiv · doi:10.48550/arxiv.1611.03530 · Cited by 2,838
Scalable and accurate deep learning with electronic health records
Rajkomar, A., Oren, E., Chen, K., Dai, A. M., Hajaj, N., Hardt, M., Liu, P. J., Liu, X., Marcus, J., Sun, M., et al. (2018) · npj Digital Medicine · doi:10.1038/s41746-018-0029-1 · Cited by 2,648 · cited 1×
“…DL methods have been used in the predictive scenarios related to quality of care, and clinical outcomes where large neural networks were used as function estimators in place of classical predictive models, with reported performance better than the state-of-the-art, classical model approaches 9–11 .”
MethodScalable and accurate deep learning with electronic health records.
Rajkomar, A., Oren, E., Chen, K., Dai, A. M., Hajaj, N., Hardt, M., Liu, P. J., Liu, X., Marcus, J., Sun, M., et al. (2018) · eScholarship (California Digital Library) · Cited by 2,242
(2010) · doi:10.5555/1756006.1859921 · Cited by 1,924
Verification and Validation in Scientific Computing
Oberkampf, W. L., & Roy, C. J. (2010) · doi:10.1017/cbo9780511760396 · Cited by 1,556
The mystery of missing heritability: Genetic interactions create phantom heritability
Zuk, O., Hechter, E., Sunyaev, S. R., & Lander, E. S. (2012) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1119675109 · Cited by 1,542
Can machine-learning improve cardiovascular risk prediction using routine clinical data?
Weng, S. F., Reps, J., Kai, J., Garibaldi, J. M., & Qureshi, N. (2017) · PLOS ONE · doi:10.1371/journal.pone.0174944 · Cited by 1,346
Cited by
Sorted by citations
Items 401–500 of 583
Target Detection on Hyperspectral Images Using MCMC and VI Trained Bayesian Neural Networks
Ries, D., Adams, J., & Zollweg, J. (2022) · IEEE Aerospace Conference (AERO) · doi:10.1109/aero53065.2022.9843344 · Cited by 1
“Uncertainty quantification (UQ) of model predictions is becoming a necessity in high consequence problems [4], [5].”
BackgroundEfficient Uncertainty Quantification for Under-Constraint Prediction Following Learning Using MCMC
Roy, G., & Chakrabarty, D. (2023) · Communications in computer and information science · doi:10.1007/978-981-99-1639-9_23 · Cited by 1
Demystifying Brain Tumour Segmentation Networks: Interpretability and Uncertainty AnalysisPreprint
Natekar, P., Kori, A., & Krishnamurthi, G. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.01498 · Cited by 1
“Begoli et al. (2019) discuss the need for uncertainty in machine-assisted medical decision making and the challenges that we might face in this context.”
BackgroundAccurate and Reliable Forecasting using Stochastic Differential EquationsPreprint
Cui, P., Deng, Z., Hu, W., & Zhu, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.15041 · Cited by 1
“, 2020), medical diagnosis (Begoli et al., 2019) and business demand forecast (Zhu & Laptev, 2017).”
Mervin, L., Trapotsi, M., Afzal, A. M., Barrett, I. P., Bender, A., & Engkvist, O. (2021) · ChemRxiv · doi:10.26434/chemrxiv.14544291 · Cited by 1
Bayesian Meta-Learning Through Variational Gaussian ProcessesPreprint
Myers, V., & Sardana, N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.11044 · Cited by 1
Counterfactual explanations for machine learning models on heterogeneous data
Wang, Y. (2023) · doi:10.32657/10356/169968 · Cited by 1
Yadav, S., & Subbian, V. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.5023458 · Cited by 1
A Decade of Deep Learning-based Biomedical Image SegmentationPreprint
Yu, S., Wang, H., Wang, N., Chen, S., Wu, J., Yuan, Z., Qi, T., Zhou, Z., Xia, F., Ma, J., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.04.27.721127 · Cited by 1
“Beyond dataset design, current foundation models still largely operate as black boxes and typically lack intrinsic mechanisms for uncertainty quantification [136].”
Tran, K. D. (2026) · medRxiv · doi:10.64898/2026.08.25.26361304 · Cited by 1
Steinmetz, P., Frouin, F., Morard, V., & Buvat, I. (2026) · doi:10.64898/2026.05.04.26350496
A Bayesian Approach to Modelling Longitudinal Data in Electronic Health RecordsPreprint
Bellot, A., & van der Schaar, M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.09086
“Moreover, the sparsity of recorded information will tend to lead to substantial uncertainty about model predictions which needs to be captured for reliable inference in practice [3].”
BackgroundSafeML: Safety Monitoring of Machine Learning Classifiers through Statistical Difference MeasurePreprint
Aslansefat, K., Sorokos, I., Whiting, D., Kolagari, R. T., & Papadopoulos, Y. I. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2005.13166
“For instance, ML in medicine by [1, 2, 3], in autonomous systems e.”
BackgroundImplicit Variational Inference: the Parameter and the Predictor Space.Preprint
Pequignot, Y., Alain, M., Dallaire, P., Yeganehparast, A., Germain, P., Desharnais, J., & Laviolette, F. (2020) · arXiv (Cornell University)
Miriyala, S. S., Banerjee, R., & Mitra, K. (2020) · IEEE Symposium Series on Computational Intelligence (SSCI) · doi:10.1109/ssci47803.2020.9308325
“However, as detailed in this recent Nature© article [10], which describes the necessity for uncertainty quantification while using Artificial Intelligence (AI) and Machine Learning (ML) based models such as ANNs, there is strong need to ensure that the uncertainties and inaccuracies arising from surrogate models be mi…”
BackgroundTailored Bayes: a risk modelling framework under unequal misclassification costsPreprint
Karapanagiotis, S., Benedetto, U., Mukherjee, S., Kirk, P., & Newcombe, P. J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.01822
“Quantification of uncertainty is critically important, especially in healthcare applications (Begoli et al., 2019; Kompa et al., 2021).”
BackgroundRobust data-driven uncertainty quantification in water cycle extreme predictions
(AI4ESP), A. I. F. E. S. P., Anderson, G. J., Pan, B., Gonçalves, A. R., Lucas, D. D., Terai, C. R., Bonfils, C., & Lee, J. (2021) · doi:10.2172/1769775
“The crucial question is how to capture all sources of uncertainty in a systematic way in order for the predictions from the AI models to be maximally leveraged by decision makers [1].”
BackgroundWhat classifiers know what they don't?Preprint
Belghazi, M. I., & López-Paz, D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2107.06217
이동헌 (2020) · Seoul National University Open Repository (Seoul National University)
Gradient Perturbation-based Efficient Deep Ensembles
Chandak, A., Kar, P., & Rai, P. (2023) · Joint International Conference on Data Science & Management of Data (CODS-COMAD) · doi:10.1145/3570991.3571063
“Obtaining wellcalibrated uncertainty estimates can be as important as having high predictive performance in specific domains such as autonomous vehicles [3] and healthcare [4] as over-confident, wrong predictions can have hazardous consequences in these applications.”
BackgroundGroup Bias and the Complexity/Accuracy Tradeoff in Machine Learning-Based Trauma Triage Models
Phillips, K., Brown, K. E., Talbert, S., & Talbert, D. A. (2023) · Proceedings of the ... International Florida Artificial Intelligence Research Society Conference · doi:10.32473/flairs.36.133358
“Understandabilty In medicine, there is often a logical reticence to trust an opaque computer-based tool for decisions that impact patient safety (Begoli, Bhattacharya, and Kusnezov 2019).”
BackgroundGaussian process for calibration and control of GlueX Central Drift Chamber
McSpadden, H., Jeske, T., Britton, T., Lawrence, D., Kalra, N., & Jarvis, N. S. (2022) · doi:10.2172/1998544
Decision Analysis in Stochastic Sociocultural Systems
Sliva, A. L., Borgonovo, E., Levis, A. H., Pawlenok, C., & Plaspohl, N. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-43185-2_11
Becker, J., José, E. S., Oladzad-Abbasabady, N., & Batta, R. (2023) · doi:10.1287/educ.2023.0262
Artificial intelligence and deep learning in molecular testing
Levy, J., & Vaickus, L. J. (2023) · Elsevier eBooks · doi:10.1016/b978-0-12-822824-1.00002-x
Personalized Exercise Program Design with Machine Learning in Sensor Networks
Lu, Y. (2023) · Scalable Computing Practice and Experience · doi:10.12694/scpe.v24i4.2440 · Cited by 2
“The use of sensors in machine learning helps to record real-time data and can be customized based on the requirements of the individual to perform the exercise [6].”
MethodSupplement on the Impact of Artificial Intelligence on Uncertainty
Arend, R. J. (2024) · Springer eBooks · doi:10.1007/978-3-031-48553-4_20
Medical follow-up optimization: A Monte-Carlo planning strategyPreprint
de Saporta, B., d’Argenlieu, A. T., Sabbadin, R., & Cleynen, A. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2401.03972
Wang, Q., Qin, D., Kuang, G., Li, R., & Nie, B. (2023) · IEEE International Conference on Intelligent Transportation Systems (ITSC) · doi:10.1109/itsc57777.2023.10422494
“Uncertainty quantification is regarded as a powerful tool to improve robustness and underpins many safety-critical predictions and decisions [13, 14].”
BackgroundUncertainty Quantification for Eosinophil Segmentation
Lin, K., Brown, D. E., Syed, S., & Greene, A. (2023) · International Conference on Bioinformatics Research and Applications (ICBRA) · doi:10.1145/3632047.3632050
“However, applications of these models in deep neural networks have caused concern among scholars about their inability to measure uncertainty [3].”
BackgroundAssessing and Addressing Model Trustworthiness Trade-offs in Trauma Triage
Talbert, D. A., Phillips, K., Brown, K. E., & Talbert, S. (2024) · International Journal of Artificial Intelligence Tools · doi:10.1142/s0218213024600078
A Study of Flow Measurement Uncertainty Quantification Based on Sample Chunked Bayesian InferencePreprint
Lu, Z., Zhu, Y., Zhou, X., Yu, J., Dong, C., & Wang, H. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4869670
1-7. Technical Group on Media Engineering
Murakami, K., Haseyama, M., Tagawa, N., Tarashima, S., & Ishii, D. (2020) · The Journal of The Institute of Image Information and Television Engineers · doi:10.3169/itej.74.779
Korsunsky, N. A., Reutov, D. K., Malyshev, A. A., & Томакова, Р. А. (2024) · Proceedings of the Southwest State University Series IT Management Computer Science Computer Engineering Medical Equipment Engineering · doi:10.21869/2223-1536-2024-14-2-126-141
Yao, N., Tian, Y., das Neves, D. G., Chen, Z., Mesquita, C. T., Martins, W. D. A., dos Santos, A. A. S. M. D., Li, Y., Han, C., Zhu, F., et al. (2024) · Kardiologiia · doi:10.18087/cardio.2024.9.n2685
Efficient Scaling of Bayesian Neural Networks
Epifano, J. R., Duong, T. Q., Ramachandran, R. P., & Rasool, G. (2024) · IEEE Access · doi:10.1109/access.2024.3479913
“Uncertainty quantification is becoming increasingly important in high-stakes decision-making, as establishing trust between users and models is crucial for the widespread adoption of such models [24].”
BackgroundPathoSpotter: Angewandte Rechnerintelligenz in der Nephropathologie
de Oliveira, L. R., Chagas, P., Duarte, A. A., Calumby, R. T., Santos, E., Ângelo, M. F., & dos Santos, W. (2024) · Springer eBooks · doi:10.1007/978-3-031-65236-3_16
Developing Tools and Assessments to Evaluate Healthcare Professionals' AI Competencies
Torres-Zambrano, G. M. (2024) · Advances in medical technologies and clinical practice book series · doi:10.4018/979-8-3693-4422-4.ch005
Explaining Uncertainty in AI for Clinical Decision Support Systems
Heremans, E. R. M., & Vos, M. D. (2024) · Communications in computer and information science · doi:10.1007/978-3-031-74627-7_33
Jang, J., Na, D., & Oh, K. (2025) · Current Bioinformatics · doi:10.2174/0115748936355551241220190451
X-ray Image Analysis for Aided Detection of COVID-19
Sharma, A., Barua, T., Garg, P., Saxena, R., Goyel, P., & Parmar, V. S. (2024) · International Conference on Intelligent Computing and Emerging Communication Technologies (ICEC) · doi:10.1109/icec59683.2024.10837310
Interpretability of Uncertainty: Exploring Cortical Lesion Segmentation in Multiple Sclerosis
Molchanova, N., Cagol, A., Gordaliza, P. M., Ocampo‐Pineda, M., Lu, P., Weigel, M., Chen, X., Depeursinge, A., Granziera, C., Müller, H., et al. (2025) · Lecture notes in computer science · doi:10.1007/978-3-031-77610-6_9 · Cited by 1
张, 倩. (2025) · Advances in Clinical Medicine · doi:10.12677/acm.2025.153839
Korsunsky, N. A., Томакова, Р. А., & Brezhnev, A. V. (2025) · Proceedings of the Southwest State University Series IT Management Computer Science Computer Engineering Medical Equipment Engineering · doi:10.21869/2223-1536-2025-15-1-117-130
Recent Advances in Medical Ethics
Pradhan, P. K. (2025) · IntechOpen eBooks · doi:10.5772/intechopen.1009364
Wabina, R. S., Saowaprut, P., Yang, J., & Pitos, C. W. (2025) · International Journal of Imaging Systems and Technology · doi:10.1002/ima.70131
“Such accumulation of errors poses significant challenges to the reliability of downstream analyses, particularly in high-stakes domains like clinical decision-making [12, 13].”
Gajbhiye, A. S., Gurbade, V. V., & Puri, C. (2025) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-5801-7.ch008
Beyond Correlation: Quantum-Causal AI for Medical Diagnosis (Preprint)Preprint
Sundararajan, P. P., Ramesh, S. S. S. R., Asha, R., & Kavitha, G. (2025) · doi:10.2196/preprints.81678
Advancing Early Detection of Chronic Obstructive Pulmonary Disease Using Generative AI
Hathaway, Q. A., & Singh, Y. (2025) · Radiology Artificial Intelligence · doi:10.1148/ryai.250555
Factors Influencing the Effectiveness of AI-Assisted Decision-Making in Medicine: A Scoping ReviewPreprint
Jackson, N. J., Brown, K. E., Miller, R., Murrow, M., Cauley, M. R., Collins, B. X., Novak, L. L., Benda, N. C., & Ancker, J. S. (2025) · medRxiv · doi:10.1101/2025.09.02.25334863
Lin, C., Chen, Z. L., Wu, C., Wu, T., & Ma, C. (2025) · IEEE Access · doi:10.1109/access.2025.3612957
“These limitations necessitate frameworks that integrate uncertainty quantification to provide calibrated predictions and enable reliable decision-making [13], [14], [15], [16], [17].”
Wen, C., Li, C., & Li, J. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.130019
Han, T., Shin, J., Lee, J. H., & Gu, K. (2025) · doi:10.2196/preprints.86498
Poslon, L., & Čartolovni, A. (2025) · Springer eBooks · doi:10.1007/978-3-031-91379-2_3
MediFusionNet : A Novel Architecture for Multimodal Medical Image Analysis
Khan, M. A., Park, H., Yamkhin, D., & Paek, S. (2025) · International Journal of Imaging Systems and Technology · doi:10.1002/ima.70246
“Uncertainty quantification is essential for responsible clinical deployment of AI systems [6].”
Bayesian Convolution Neural Network with Spatially Correlated Priors
Hwang, T., Muthukumar, L. B. G., & Darak, Y. G. (2025) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn64981.2025.11228759
Korsunsky, N. A., Томакова, Р. А., & Starkov, V. A. (2025) · Proceedings of Southwest State University · doi:10.21869/2223-1560-2025-29-3-124-136
crispAIPE: Probabilistic Modelling of Prime Editing Variant Correction EfficiencyPreprint
Özden, F., Lu, P., & Mináry, P. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2025.12.07.692852
“Uncertainty quantification is valuable wherever prediction reliability drives decisions [6, 18], including CRISPR off-target assessment [31] and guide selection [36]; in prime editing the added challenge is that the three competing outcomes are compositional and must sum to unity, a constraint…”
Dolezal, J. M., Kochanny, S. E., Dyer, E., Ramesh, S., Srisuwananukorn, A., Sacco, M., Howard, F. M., Li, A., Mohan, P., & Pearson, A. T. (2024) · Figshare · doi:10.6084/m9.figshare.26698715
Wang, D., Yu, J., Chen, L., Li, X., Jiang, H., Chen, K., Zheng, M., & Luo, X. (2021) · Figshare · doi:10.6084/m9.figshare.16650122
Mervin, L., Trapotsi, M.-A., Afzal, A. M., Barrett, I. P., Bender, A., & Engkvist, O. (2021) · Figshare · doi:10.6084/m9.figshare.15569340
Mervin, L., Trapotsi, M.-A., Afzal, A. M., Barrett, I. P., Bender, A., & Engkvist, O. (2021) · Figshare · doi:10.6084/m9.figshare.15569352
Beil, M., van Heerden, P. V., de Lange, D. W., Szczeklik, W., Leaver, S. K., Guidet, B., Flaatten, H., Jung, C., Sviri, S., & Leo Joskowicz, B. (2024) · Figshare · doi:10.6084/m9.figshare.26559701
AI-Aided Decision-Making in Education: Uncertainty, Explainability, and Human Responsibility
Kieser, F., Tschisgale, P., & Wulff, P. (2025) · Signals and communication technology · doi:10.1007/978-3-031-96855-6_2
Shah-Mohammadi, F., Millar, A., Facelli, J., & Gouripeddi, R. (2025) · medRxiv · doi:10.64898/2025.12.14.25342221
Zgodic, A., Bai, R., Zhang, J., Wang, Y. Q., Rorden, C., & McLain, A. C. (2025) · Computational Statistics · doi:10.1007/s00180-025-01699-y
“Accurately assessing the predictive uncertainty is essential to the acceptance and effectiveness of models in risk-sensitive healthcare settings (Banerji et al. 2023; Begoli et al. 2019).”
InfluentialSquires, S., Kuling, G., Evans, D. G., Martel, A. L., & Astley, S. (2026) · Biomedical Physics & Engineering Express · doi:10.1088/2057-1976/ae39e2
Extrinsic Trust as a Contractual Framework for Accountable AI in Health Care: Viewpoint
Kelly, A. (2026) · Journal of Medical Internet Research · doi:10.2196/83903 · Cited by 1
“Without external validation, especially under real-world conditions where data distributions may shift, clinicians may withhold trust or misplace it, leading to underuse or overreliance [12,14,18].”
Threshold and Calibration Effects in Medical Machine Learning
Özcan, H. (2025) · International Conference on Algorithms, Computing and Artificial Intelligence (ACAI) · doi:10.1109/acai68217.2025.11406486
“These shortcomings undermine clinical trust and patient safety, reinforcing calls for explicit uncertainty awareness in decision support systems [12].”
Security and risk governance in AI-driven medical imaging systems: An uncertainty-aware framework
Ahmad, S., Kumar, K., Prakash, A., & Shahnawaz (2026) · EDPACS · doi:10.1080/07366981.2026.2640131
Mining hidden layers in CNN for certainty prediction in medical imaging
Hirimutugoda, Y. M., Sliva, T. P., & Wagarachchi, N. M. (2026) · Neural Computing and Applications · doi:10.1007/s00521-025-11840-z
FedBNR: A Fully Global Federated Gaussian Process
Yu, H., Guo, K., Karami, M., Zhang, G., Li, H., Chen, X., & Poupart, P. (2026) · Machine Learning · doi:10.1007/s10994-025-06936-5
“In machine-assisted decision making (Begoli et al., 2019), it provides the trust needed to inform decisions.”
Shchetinin, E. Y., & Шевчук, А. А. (2026) · Modelirovanie, optimizaciâ i informacionnye tehnologii. · doi:10.26102/2310-6018/2026.54.3.009
Age and sex estimation using post-mortem CT scout view
Provost, C., Richard, M.-E., Delabarde, T., Hmeydia, G., Berre, A. L., Hassen, W. B., Naggara, O. N., Hamza, L., Ludes, B., Oppenheim, C., et al. (2026) · International Journal of Legal Medicine · doi:10.1007/s00414-026-03772-7
“Including a measure and visu - alization of uncertainty is essential for modern AI decision-making tools [46].”
Personalizing anesthetic management in trauma patients using machine learning
James, A., & Moyer, J.-D. (2026) · Current Opinion in Anaesthesiology · doi:10.1097/aco.0000000000001637
Deep Probabilistic Forecasting for Market Risks
Zhang, Z., Zhang, K., Yan, X., Yang, S., & Zhang, Y. (2026) · Springer eBooks · doi:10.1007/978-981-95-3125-7_7
Bayesian neuromorphic computing based on resistive memory
Dalgaty, T. (2020) · theses.fr (ABES) · doi:10.70675/e5b4f3e2z44a5z4c41zbac6z4c933e86b101
Statistical learning and causal inference for energy production
Acharki, N. (2022) · theses.fr (ABES) · doi:10.70675/fc461b55z6d2cz41b8z9c2az43774adf2b0c
Uncertainty-Calibrated Decision Thresholds in Bayesian AI Systems: A Governance-Oriented Framework
Raham, T. F. (2026) · Asian Journal of Probability and Statistics · doi:10.9734/ajpas/2026/v28i4887
Sy, O. O., Islam, A., van Beurden, M., Crow, K., Woods, S., & Guerra, D. J. D. (2026) · International Journal for Uncertainty Quantification · doi:10.1615/int.j.uncertaintyquantification.2026060421
Harada, Y. (2026) · Cureus · doi:10.7759/cureus.107548
“This is conceptually related to selective prediction and to arguments that clinical AI systems should communicate or act on uncertainty when patient-level risk is high [17,22,23].”
Mahak, Singh, S., Sankhayadhar, N., Khan, I. J., & Tripathi, A. (2026) · De Gruyter eBooks · doi:10.1515/9783112226223-001
Meta Relation-Aware Metric Learning Framework for Few-Shot Uncertain Knowledge Graph Completion
Wang, X. G., Yang, K., Liu, J., Li, X., Hua, R., Xu, K., Shu, Z., & Zhou, X. (2026) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2026.3692238
Composite Certainty: Addressing Metric Degeneracy in Parameter Inference for Model-Based DiagnosticsPreprint
Koshe, A., Tehrani, E. S., Jalaleddini, K., & Motallebzadeh, H. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.05.09.724027
Sajid, M. Z., & Hamid, M. F. (2026) · Information Sciences · doi:10.1016/j.ins.2026.123620
Fan, C., Zhang, J., Wang, S., Guo, Y., & Fu, R. (2026) · Frontiers in Plant Science · doi:10.3389/fpls.2026.1808419
Takayidza, C., Musara, M., & Takayidza, C. (2026) · Research Square · doi:10.21203/rs.3.rs-9732599/v1
Li, F., Furutani, K. M., & Beltran, C. J. (2026) · Tomography · doi:10.3390/tomography12050066
“Equally important is the capacity of an AI algorithm to provide a confidence level or uncertainty measure, particularly for deployment in healthcare settings where erroneous predictions carry significant harm [259–262].”
Min, J., Vishnyakova, O., Brooks-Wilson, A., & Elliott, L. T. (2026) · GeroScience · doi:10.1007/s11357-026-02280-1
Uncertainty-guided model learning for trustworthy medical image segmentation
Ye, Q., Guo, L., & Wu, Q. (2026) · Medical & Biological Engineering & Computing · doi:10.1007/s11517-026-03591-3
“Although these models exhibit strong performance, they are still prone to produce unreliable predictions and are questioned by clinicians [9–11].”
Shater, K., Gerhards, C., Evliyaoğlu, O., Nittka, S., & Fischer, A. (2026) · AI in Medicine · doi:10.3390/aimed1020013
Schatzkästchen oder Büchse der Pandora? – Klinische Ethik und Künstliche Intelligenz
Beil, M., Janssens, U., & Fuest, K. (2026) · Springer eBooks · doi:10.1007/978-3-662-72378-4_16
A Unified Framework for Reliable Survival Analysis
Massoua, A. B., Bouguessa, M., & Diallo, A. B. (2026) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2026.3697686
Metadata
- Authors
- Begoli, E., Bhattacharya, T., & Kusnezov, D.
- Affiliations
- Oak Ridge National Laboratory (US)
- United States Department of Energy (US)
- Corresponding author
- Begoli, Edmon
- ORCID
- Begoli, Edmon
- Venue
- Nature Machine Intelligence
- Published
- 2019-01-07
- Volume
- 1
- Issue
- 1
- Pages
- 20-23
- ISSN
- 2522-5839 (ISSN-L)
- OpenAlex
- W2904422658
- Semantic Scholar
- a73eef4e0ec8e49be1c0e6abb04cc1967a681c50
- Cited by
- 597
- References
- 22
- Citations by year
- 2026: 158 · 2025: 102 · 2024: 73 · 2023: 94 · 2022: 72 · 2021: 47 · 2020: 33 · 2019: 16
- FWCI
- 6.47
- Type
- journal article
- Language
- en
- Open access
- green
- www.osti.gov
- Licence
- Locations
- Nature Machine Intelligence (publishedVersion)
- OSTI OAI (U.S. Department of Energy Office of Scientific and Technical Information) (submittedVersion)
- Funders
- Battelle (DE-AC05- 00OR22725, DE-AC05)
- UT-Battelle (DE-AC05, DEAC05-00OR22725, AC05-00OR22725)
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Keywords
- uncertainty quantification, artificial intelligence in healthcare, deep learning, patient safety, trust in AI, forecast uncertainty
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-09, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-09 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
