The DOI matches its Crossref record; 31 of the 31 references it cites were found in a registry or an index; 6 of the 23 sources its claims cite were checked for support.
Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency
Open accessAbstract
Machine Intelligence (MI) is rapidly becoming an important approach across biomedical discovery, clinical research, medical diagnostics/devices, and precision medicine. Such tools can uncover new possibilities for researchers, physicians, and patients, allowing them to make more informed decisions and achieve better outcomes. When deployed in healthcare settings, these approaches have the potential to enhance efficiency and effectiveness of the health research and care ecosystem, and ultimately improve quality of patient care. In response to the increased use of MI in healthcare, and issues associated when applying such approaches to clinical care settings, the National Institutes of Health (NIH) and National Center for Advancing Translational Sciences (NCATS) co-hosted a Machine Intelligence in Healthcare workshop with the National Cancer Institute (NCI) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB) on 12 July 2019. Speakers and attendees included researchers, clinicians and patients/ patient advocates, with representation from industry, academia, and federal agencies. A number of issues were addressed, including: data quality and quantity; access and use of electronic health records (EHRs); transparency and explainability of the system in contrast to the entire clinical workflow; and the impact of bias on system outputs, among other topics. This whitepaper reports on key issues associated with MI specific to applications in the healthcare field, identifies areas of improvement for MI systems in the context of healthcare, and proposes avenues and solutions for these issues, with the aim of surfacing key areas that, if appropriately addressed, could accelerate progress in the field effectively, transparently, and ethically.
Overview
CoChat Summary
TL;DR
This whitepaper reports on an NIH/NCATS co-hosted workshop held 12 July 2019 on machine intelligence (MI) in healthcare, focused on trustworthiness, explainability, usability, and transparency/fairness. It synthesizes expert perspectives on data, infrastructure, workflow, and system challenges and proposes approaches to support effective, transparent, and ethical MI adoption.
Key findings
- The workshop identified four priority, overlapping topics for healthcare MI: trustworthiness, explainability, usability, and transparency/fairness.
- Trustworthiness requires confidence across the data, system, output, and workflow, including continual surveillance and structured feedback loops after deployment; different audiences may require different validation testing.
- Explainability includes interpretability and completeness, and may not always be necessary if trust is already high, but it is especially important now because MI applications lack the historical evidence base of some accepted drugs.
- Usability barriers include EHR point-of-care settings remaining a “walled garden” due to isolated, noninteroperable implementations; proposed enablers include SMART on FHIR APIs, CDS Hooks, and SMART/HL7 FHIR bulk data export.
- Bias can enter MI systems through population transfer, malicious skewing, inadequate or narrow datasets, algorithm structure, and equipment measurement drift; suggested responses include representative datasets, bias-identification methods, and multi-site/multi-dataset validation.
Methods
- Design: a whitepaper/report from an expert workshop, not an empirical study or systematic review.
- Data or sample: the NIH National Center for Advancing Translational Sciences (NCATS) co-hosted a Machine Intelligence in Healthcare workshop with the National Cancer Institute (NCI) and the National Institute of Biomedical Imaging and Bioengineering (NIBIB) on 12 July 2019; attendees included researchers, clinicians, and patients/patient advocates from industry, academia, and federal agencies.
- Analysis: narrative synthesis of expert panel discussions on trustworthiness, explainability, usability, and transparency/fairness, reporting key issues, gaps, barriers, and proposed approaches; no quantitative analysis is reported.
Limitations
- The paper does not present a formal limitations section; it states many challenges and gaps, including evolving MI systems, “black box” complexity, data quality/quantity, EHR interoperability, bias, need for clearer definitions, and need for interdisciplinary collaboration and post-deployment monitoring.
- Evident from the text: as a workshop whitepaper, it reflects expert perspectives and consensus discussion rather than systematic review or empirical evaluation.
- Evident from the text: the proposed solutions and recommendations are conceptual and are not tested or validated within the paper.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 0 Not valid · 11 Not checked · 18 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“However, it has been shown that interoperability can be implemented successfully for large data-dependent systems, such as with the World Wide Web, and there is increasing awareness of the importance of interoperability within digital health15–17.”
§ Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency > Results > Session 3: Usability of MI in healthcare
What the sources say
- Not checked yetcites [15] Perlin, J. B. Health information technology interoperability and use for better care and evidence. JAMA 316, 1667–1668 (2016). source
- Not checked yetcites [16] Procuring Interoperability: Achieving High-Quality, Connected, and Person-Centered Care. Natl Acad. Med. https://nam.edu/procuring-interoperability-achieving-high-quality-connected-and-person-centered-care/ (2018).
- Partly supportedcites [17] Lehne, M., Sass, J., Essenwanger, A., Schepers, J. & Thun, S. Why digital medicine depends on interoperability. Npj Digit. Med. 2, 1–5 (2019). sourceExperimental — accuracy not yet measured
Our copy of Lehne et al. (2019) supports only the second half of the citing sentence: the Introduction states that 'the importance of interoperable health IT systems is increasingly acknowledged' (while awareness remains relatively low among healthcare professionals). The specific proposition attributed — that interoperability has been shown to work successfully for large data-dependent systems such as the World Wide Web — is not in our copy: searches for 'World Wide Web', web/internet standards and success-story phrasing returned no such statement. The only web-related content is the Syntactic interoperability section noting that FHIR resources 'can be accessed and exchanged using modern web technologies', and the Technical interoperability section saying that with today's digital networks and communication protocols technical interoperability is 'usually relatively straightforward' — neither presents the Web as a demonstrated large-scale interoperability success. The unsupported part of the claim is therefore the World Wide Web example/proof. (LLM, conf=0.65)
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.
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“However, it has been shown that interoperability can be implemented successfully for large data-dependent systems, such as with the World Wide Web, and there is increasing awareness of the importance of interoperability within digital health15–17.”
§ Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency > Results > Session 3: Usability of MI in healthcare
What the sources say
- Not checked yetcites [15] Perlin, J. B. Health information technology interoperability and use for better care and evidence. JAMA 316, 1667–1668 (2016). source
- Not checked yetcites [16] Procuring Interoperability: Achieving High-Quality, Connected, and Person-Centered Care. Natl Acad. Med. https://nam.edu/procuring-interoperability-achieving-high-quality-connected-and-person-centered-care/ (2018).
- Supportedcites [17] Lehne, M., Sass, J., Essenwanger, A., Schepers, J. & Thun, S. Why digital medicine depends on interoperability. Npj Digit. Med. 2, 1–5 (2019). source
The Introduction (para. 2) states outright that 'the importance of interoperable health IT systems is increasingly acknowledged,' and frames the article's argument that interoperability is indispensable for advances in digital health — exactly what the citing clause attributes. The Abstract and Conclusion and outlook reinforce this (interoperability as a prerequisite for future digital medicine, and the call to make it a prominent topic in medicine and healthcare). The paper's accompanying caveat that awareness remains relatively low among healthcare professionals does not contradict the claim of growing awareness; the separate World Wide Web clause is not what this citation cluster is offered for. (LLM, conf=0.88)
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.
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Particular issues are presented by the use of published data, which can be unvalidated or incomplete due to selective reporting26,27.”
§ Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency > Discussion > Gaps, barriers, and methodologies for implementation of MI in healthcare
What the sources say
- Could not verifycites [26] Stupple, A., Singerman, D. & Celi, L. A. The reproducibility crisis in the age of digital medicine. Npj Digit. Med. 2, 1–3 (2019). sourceExperimental — accuracy not yet measured
Our copy of Shebib et al. (2019) is a single randomized trial of a 12-week digital care program; it never discusses using published/secondary data, nor unvalidated or incomplete data due to selective reporting (searched: 'selective reporting of published data unvalidated incomplete', 'limitations of using published data sources bias', 'publication bias reporting of outcomes registry data source', 'self-reported outcomes data not validated questionnaire reliability'). Its Discussion (limitations paragraph) names only non-blinding, the remote/self-report nature of the program and lack of clinician assessment or medical-record review, and its Methods (Study outcomes) address missing-data handling (LOCF/BOCF, mixed-model, per-protocol) rather than reporting bias in published literature. Nothing here supports or contradicts the attributed point, so the claim cannot be judged against this source. (LLM, conf=0.70)
- Partly supportedcites [27] Wallach, J. D., Boyack, K. W. & Ioannidis, J. P. A. Reproducible research practices, transparency, and open access data in the biomedical literature, 2015–2017. PLOS Biol. 16, e2006930 (2018). sourceExperimental — accuracy not yet measured
The Discussion/Limitations passage (p. n/a, 'Limitations') does support the core idea that relying on published records poses particular problems: the authors' evaluation 'relied on published records' and based novelty determinations on 'the information reported by investigators', noting it is 'possible that authors may have tried to spin their manuscript as being more novel than it really is'. Other passages support incompleteness of published reporting (Results, 'Protocol and raw data availability': only 18.3% discussed publicly available data and 1.0% linked a full protocol; Discussion: conflicts-of-interest reporting 'may underestimate the true prevalence'). However, our copy nowhere attributes the unvalidated/incomplete character of published data to 'selective reporting' as such — the paper frames this as its own study limitation and hedges the spin point with 'it is possible', so the causal qualifier and the generalised factual framing of the citing sentence go beyond what this source states. (LLM, conf=0.68)
plainstates a cause- 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.
- Could not verify
Every source the paper cites for this claim was checked and supports it.
In the paper
“Introduction of bias is a pressing issue that has become a focal point of MI fairness21,22.”
§ Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency > Results > Session 4: Transparency and fairness of MI in healthcare
What the sources say
- Supportedcites [21] Obermeyer, Z., Powers, B., Vogeli, C. & Mullainathan, S. Dissecting racial bias in an algorithm used to manage the health of populations. Science 366, 447–453 (2019). source
The claim is a broad background statement about how bias enters algorithms being a pressing concern in machine-intelligence fairness, and our copy treats the topic exactly that way: the opening of the article (p. 2) states there is 'growing concern that algorithms may reproduce racial and gender disparities via the people building them or through the data used to train them,' the abstract (p. 2) concludes that 'the choice of convenient, seemingly effective proxies for ground truth can be an important source of algorithmic bias in many contexts,' and the Discussion (p. 8) calls this 'mechanism of bias ... particularly pernicious because it can arise from reasonable choices.' The dedicated 'Mechanism of bias' section (pp. 4-5) and the closing remarks (p. 2) on 'the importance, and the large opportunity, of studying algorithmic bias' reinforce that identifying the sources of bias is framed as central rather than incidental. The paper does not use the phrase 'MI fairness,' but the substance attributed to it is present. (LLM, conf=0.82)
- Supportedcites [22] Vayena, E., Blasimme, A. & Cohen, I. G. Machine learning in medicine: addressing ethical challenges. PLOS Med. 15, e1002689 (2018). source
The fairness section of our copy ('MLm development should be committed to fairness') states that poorly representative training data sets 'can introduce biases into MLm-trained algorithms' and elaborates two archetypes of bias in medical data, plus the potential to exacerbate health disparities — so the introduction of bias is exactly what the paper treats as the core issue of MLm fairness. The Abstract likewise lists 'data fairness' among the central ethical and regulatory challenges of AI in healthcare. The paraphrase's 'pressing issue / focal point' framing matches how our copy organizes and emphasizes the topic; 'MI' vs the paper's 'MLm' is a wording difference only. (LLM, conf=0.85)
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.
- Supported
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 historical data do not yet exist with MI applications in healthcare.”
§ Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency > Results > Session 2: Explainability of MI in healthcare
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: unclear — It is unclear why the paper says it holds.
11 more claims
Create a free account to see all 33 claims and to verify the 17 references still open.
- Probably not a substantial claim. 16
- It may state two separate claims as one. 8
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 cited source says what the claim attributes to it.
In the citing paper
“This divergence was encouraging and suggested a gradual shift toward patient-centered, high-quality evidence capable of building clinical trust25,26 and satisfying regulatory expectations27,28.”
From A quantitative analysis of global AI medical studies: gaps in randomized controlled trials (2026) · cites this paper as [27]
In the citing paper
“Given these considerations, explainability may be a key driver for the uptake of AI-driven CDSS in clinical practice, as trust in these systems is not yet established [22, 23].”
From Explainability for artificial intelligence in healthcare: a multidisciplinary perspective (2020) · cites this paper as [22]
In the citing paper
“Crucially, there should be a clearly identified clinical problem that is to be tackled [28], noting how ML outputs are not always actionable and hence may not necessarily help solve a problem [169].”
From Operationalising ethics in artificial intelligence for healthcare: a framework for AI developers (2023) · cites this paper as [169]
Citations
Cites
High-performance medicine: the convergence of human and artificial intelligence
Topol, E. J. (2019) · Nature Medicine · doi:10.1038/s41591-018-0300-7 · Cited by 10,450
Dissecting racial bias in an algorithm used to manage the health of populations
Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019) · Science · doi:10.1126/science.aax2342 · Cited by 7,137
Moons, K. G. M., Altman, D. G., Reitsma, J. B., Ioannidis, J. P. A., Macaskill, P., Steyerberg, E. W., Vickers, A. J., Ransohoff, D. F., & Collins, G. S. (2015) · Annals of Internal Medicine · doi:10.7326/m14-0698 · Cited by 5,147
Clinically applicable deep learning for diagnosis and referral in retinal disease
De Fauw, J., Ledsam, J. R., Romera-Paredes, B., Nikolov, S., Tomasev, N., Blackwell, S., Askham, H., Glorot, X., O’Donoghue, B., Visentin, D., et al. (2018) · Nature Medicine · doi:10.1038/s41591-018-0107-6 · Cited by 2,639
Do no harm: a roadmap for responsible machine learning for health care
Wiens, J., Saria, S., Sendak, M., Ghassemi, M., Liu, V. X., Doshi-Velez, F., Jung, K., Heller, K., Kale, D., Saeed, M., et al. (2019) · Nature Medicine · doi:10.1038/s41591-019-0548-6 · Cited by 1,266
Adversarial attacks on medical machine learning
Finlayson, S. G., Bowers, J. D., Ito, J., Zittrain, J. L., Beam, A. L., & Kohane, I. S. (2019) · Science · doi:10.1126/science.aaw4399 · Cited by 1,088
Machine learning in medicine: Addressing ethical challenges
Vayena, E., Blasimme, A., & Cohen, I. G. (2018) · PLOS Medicine · doi:10.1371/journal.pmed.1002689 · Cited by 912
SMART on FHIR: a standards-based, interoperable apps platform for electronic health records
Mandel, J. C., Kreda, D. A., Mandl, K. D., Kohane, I. S., & Ramoni, R. B. (2016) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocv189 · Cited by 829
Self-citationUnequal Treatment—Confronting Racial and Ethnic Disparities in Healthcare
Osundeko, O. (2004) · PubMed Central · Cited by 797
Rationalizing Neural Predictions
Lei, T., Barzilay, R., & Jaakkola, T. (2016) · Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/d16-1011 · Cited by 777
Research works printed but not linked: 1
- Schulam, P. & Saria, S. Can you trust this prediction? Auditing pointwise reliability after learning. PMLR. 89, 1022–1031 (2019).
Cited by
Rajpurkar, P., Chen, E., Banerjee, O., & Topol, E. J. (2022) · Nature Medicine · doi:10.1038/s41591-021-01614-0 · Cited by 3,212
“For example, it is useful for AI systems to be reliable, convenient to use and easy to integrate into clinical workflows 98 .”
MethodExplainability for artificial intelligence in healthcare: a multidisciplinary perspective
the Precise4Q consortium, Amann, J., Blasimme, A., Vayena, E., Frey, D., & Madai, V. I. (2020) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-020-01332-6 · Cited by 2,203
The false hope of current approaches to explainable artificial intelligence in health care
Ghassemi, M., Oakden-Rayner, L., & Beam, A. L. (2021) · The Lancet Digital Health · doi:10.1016/s2589-7500(21)00208-9 · Cited by 1,724
Do as AI say: susceptibility in deployment of clinical decision-aids
Gaube, S., Suresh, H., Raue, M., Merritt, A., Berkowitz, S. J., Lermer, E., Coughlin, J. F., Guttag, J. V., Colak, E., & Ghassemi, M. (2021) · npj Digital Medicine · doi:10.1038/s41746-021-00385-9 · Cited by 543
Artificial intelligence for clinical oncology
Kann, B. H., Hosny, A., & Aerts, H. J. W. L. (2021) · Cancer Cell · doi:10.1016/j.ccell.2021.04.002 · Cited by 370
Metadata
- Authors
- Cutillo, C. M., Sharma, K. R., Foschini, L., Kundu, S., Mackintosh, M., Mandl, K. D., MI in Healthcare Workshop Working Group, Beck, T., Collier, E., Colvis, C., Gersing, K., Gordon, V., Jensen, R., Shabestari, B., & Southall, N.
- Affiliations
- National Institutes of Health (US)
- Evidation Health (United States) (US)
- Johns Hopkins University (US)
- The Alan Turing Institute (GB)
- Boston Children's Hospital (US)
- Corresponding author
- Cutillo, Christine M.
- Mandl, Kenneth D.
- ORCID
- Venue
- npj Digital Medicine
- Published
- 2020-03-26
- Volume
- 3
- Issue
- 1
- Pages
- 47-47
- ISSN
- 2398-6352 (ISSN-L)
- PMC
- PMC7099019
- OpenAlex
- W3013443134
- Semantic Scholar
- 74d405164554f1432c66c39210d8fb4331aeee54
- Cited by
- 337
- References
- 30
- Citations by year
- 2026: 48 · 2025: 69 · 2024: 71 · 2023: 63 · 2022: 52 · 2021: 30 · 2020: 4
- FWCI
- 10.97
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- npj Digital Medicine (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Vanderbilt University
- Massachusetts Institute of Technology
- Johns Hopkins University
- Vanderbilt University Medical Center
- York University
- University of Toronto
- National Institutes of Health
- Agency for Healthcare Research and Quality
- National Institute of Biomedical Imaging and Bioengineering
- National Center for Advancing Translational Sciences
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- trustworthiness, explainability, usability, transparency, algorithmic bias, electronic health records, data quality, clinical decision support, AI ethics, quality of care, precision medicine
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency. https://app.cochat.ai/papers/10.1038/s41746-020-0254-2
- The paper
- Cutillo, C. M., Sharma, K. R., Foschini, L., Kundu, S., Mackintosh, M., Mandl, K. D., MI in Healthcare Workshop Working Group, Beck, T., Collier, E., Colvis, C., Gersing, K., Gordon, V., Jensen, R., Shabestari, B., & Southall, N. (2020). Machine intelligence in healthcare—perspectives on trustworthiness, explainability, usability, and transparency. npj Digital Medicine. https://doi.org/10.1038/s41746-020-0254-2
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.