Machine learning-based clinical decision support using laboratory data

Probably not a substantial claim: may not pass S3 (load-bearing).

A tuning dataset can help improve model performance by adjusting parameters and configurations.

“The inclusion of a tuning dataset, while dependent on the methodology applied, can be valuable in optimizing the model's performance by fine-tuning its parameters and configurations [5].”

• cites [5] Faes, L, Liu, X, Wagner, SK, Fu, DJ, Balaskas, K, Sim, DA, et al.. A clinician’s guide to artificial intelligence: how to critically appraise machine learning studies. Transl Vis Sci Technol 2020;9:7. https://doi.org/10.1167/tvst.9.2.7. sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In two well-known contests, AI systems beat human champions: Deep Blue against Garry Kasparov and AlphaGo against Lee Se-dol.

“Two notable clashes, namely Deep Blue vs. Garry Kasparov and AlphaGo against Lee Se-dol, have served as compelling demonstrations of AI's superiority in challenging cognitive tasks [1].”

• cites [1] Granter, SR, Beck, AH, Papke, DJJr. AlphaGo, deep learning, and the future of the human microscopist. Arch Pathol Lab Med 2017;141:619–21. https://doi.org/10.5858/arpa.2016-0471-ed. sourceSupported

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Integrated diagnostics, which combines radiology, pathology, and laboratory medicine with advanced information technology, has great promise for changing disease diagnosis and therapeutic interventions.

“Integrated diagnostics, which entails the integration of radiology, pathology, and laboratory medicine with advanced information technology, holds immense promise in transforming the landscape of disease diagnosis and therapeutic interventions [21].”

• cites [21] Lippi, G, Plebani, M. Integrated diagnostics: the future of laboratory medicine? Biochem Med 2020;30:010501. https://doi.org/10.11613/bm.2020.010501. sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

In vitro diagnostics underpin about two-thirds of clinical decisions.

“A significant proportion, 66 %, of clinical decisionmaking is based on in vitro diagnostics (IVD) [15].”

• cites [15] Rohr, UP, Binder, C, Dieterle, T, Giusti, F, Messina, CG, Toerien, E, et al.. The value of in vitro diagnostic testing in medical practice: a status report. PLoS One 2016;11:e0149856. https://doi.org/10.1371/journal.pone.0149856. sourceSupported

May combine two claims.

The SVM model achieved 85.11% accuracy for reactive lymphoid cells, 82% for myeloid blast cells, and 73.97% for lymphoid blast cells.

“Notably, the SVM model exhibited high accuracy in distinguishing reactive lymphoid cells (85.11 %) and myeloid blast cells (82 %), although the accuracy for lymphoid blast cells was relatively lower (73.97 %).”

• cites [33] Bigorra, L, Merino, A, Alférez, S, Rodellar, J. Feature analysis and automatic identification of leukemic lineage blast cells and reactive lymphoid cells from peripheral blood cell images. J Clin Lab Anal 2017;31:1–9. https://doi.org/10.1002/jcla.22024. sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Techniques available for explainable AI are still maturing and may not be sufficient to address all the complexities inherent in advanced machine learning models.

“Techniques available for XAI are still maturing and may not be sufficient to address all the complexities inherent in advanced ML models [127].”

• cites [127] Ghassemi, M, Oakden-Rayner, L, Beam, AL. The false hope of current approaches to explainable artificial intelligence in health care. Lancet Digit Health 2021;3:e745–50. https://doi.org/10.1016/s2589-7500(21)00208-9. sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

The FAIR data principles support high data management and stewardship standards, ensuring that data used for training and validating machine learning models are well-curated, standardized, and transparent.

“The FAIR data principles support for a high data management and stewardship standard, ensuring that the data used for training and validating ML models are well-curated, standardized, and transparent [114, 115].”

• cites [114] Bellamy, RKE, Dey, K, Hind, M, Hoffman, SC, Houde, S, Kannan, K, et al.. AI Fairness 360: an extensible toolkit for detecting and mitigating algorithmic bias. IBM J Res Dev 2019;63:4:1–4:15. https://doi.org/10.1147/jrd.2019.2942287. sourceCould not verify

  The Abstract and Introduction (p. 1) describe a toolkit for detecting and mitigating algorithmic bias, not the FAIR data principles. Searches for FAIR data stewardship and for curated, standardized, transparent training and validation data found no passage addressing the attributed claim. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [115] Wilkinson, MD, Dumontier, M, Aalbersberg, IJ, Appleton, G, Axton, M, Baak, A, et al.. The FAIR Guiding Principles for scientific data management and stewardship. Sci Data 2016;3:160018. https://doi.org/10.1038/sdata.2016.18. sourcePartly supported

  The section “FAIRness is a prerequisite for proper data management and data stewardship” says the principles guide good data management and stewardship. The paper also describes standards-based metadata and curation in “Examples of FAIRness, and the resulting value-added.” It does not address datasets for training or validating ML models, or say that applying FAIR principles ensures those datasets are well-curated, standardized, and transparent. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Model-agnostic explainers exhibit greater flexibility than model-specific explainers and can be used to interpret any machine learning model.

“Therefore model-agnostic explainers exhibit greater flexibility than model-specific explainers and can be used to interpret any ML model [118, 121, 122].”

• cites [118] Barredo Arrieta, A, Díaz-Rodríguez, N, Del Ser, J, Bennetot, A, Tabik, S, Barbado, A, et al.. Explainable Artificial Intelligence (XAI): concepts, taxonomies, opportunities and challenges toward responsible AI. Inf Fusion 2020;58:82–115. https://doi.org/10.1016/j.inffus.2019.12.012. sourceCould not verify

  The Abstract describes a taxonomy of explainability methods but does not compare model-agnostic and model-specific explainers. Searches for “model-agnostic model-specific flexibility any model” and “agnostic” returned only the abstract and repository metadata, so the available passages do not establish the claimed flexibility or applicability to any ML model. (LLM, conf=0.72)

  Experimental — accuracy not yet measured
• cites [121] Ali, S, Abuhmed, T, El-Sappagh, S, Muhammad, K, Alonso-Moral, JM, Confalonieri, R, et al.. Explainable Artificial Intelligence (XAI): what we know and what is left to attain trustworthy artificial intelligence. Inf Fusion 2023;99:101805. https://doi.org/10.1016/j.inffus.2023.101805. sourceCould not verify

  The available passages in §1.1 discuss post-hoc explainers and interpretable-by-design approaches, while §1.2 discusses how explanation methods are evaluated. Searches for “model-agnostic,” “model-specific,” and applicability to any ML model found no passage comparing their flexibility or stating that model-agnostic explainers can interpret any model. (LLM, conf=0.76)

  Experimental — accuracy not yet measured
• cites [122] Markus, AF, Kors, JA, Rijnbeek, PR. The role of explainability in creating trustworthy artificial intelligence for health care: a comprehensive survey of the terminology, design choices, and evaluation strategies. J Biomed Inf 2021;113:103655. https://doi.org/10.1016/j.jbi.2020.103655. sourceSupported

May combine two claims. The quote only partly states this claim.

The reviewed studies demonstrated the efficacy of machine learning algorithms in several diagnostic applications, including sepsis, cancer, diabetes, gestational disorders, LDL cholesterol estimation, infections, benign and malignant lesions, genetic disorders, autoimmune diseases, and hematological malignancies.

“The reviewed studies demonstrated the efficacy of ML algorithms in several diagnostic applications.”

• Also stated in: § Machine learning in the post-analytical phase · plain

  “Predictive models were developed for diseases such as sepsis [41], cancer [78], diabetes, and gestational disorders [62].”

• Also stated in: § Machine learning in the post-analytical phase · plain

  “ML algorithms were also used for estimating LDL cholesterol levels [106], identifying specific infections [88], and distinguishing between benign and malignant lesions [58].”

• cites [41] Aguirre, U, Urrechaga, E. Diagnostic performance of machine learning models using cell population data for the detection of sepsis: a comparative study. Clin Chem Lab Med 2023;61:356–65. https://doi.org/10.1515/cclm-2022-0713. sourceNot checked — no full text of the source
• cites [78] Reix, N, Lodi, M, Jankowski, S, Molière, S, Luporsi, E, Leblanc, S, et al.. A novel machine learning-derived decision tree including uPA/PAI-1 for breast cancer care. Clin Chem Lab Med 2019;57:901–10. https://doi.org/10.1515/cclm-2018-1065. sourceNot checked — no full text of the source
• cites [62] Hu, Z, Zhang, M. Establishment of clinical diagnostic models using glucose, lipid, and urinary polypeptides in gestational diabetes mellitus. J Clin Lab Anal 2021;35:e23833. https://doi.org/10.1002/jcla.23833. sourceSupported

Classified as probably not a substantial claim. May combine two claims.

Machine learning can augment human performance and reduce interobserver variability in clinical decision support.

“An optimal approach to CDSS implementation would be to use ML to augment human performance and decrease interobserver variability [138], rather than replacing them entirely.”

• cites [138] Vasey, B, Ursprung, S, Beddoe, B, Taylor, EH, Marlow, N, Bilbro, N, et al.. Association of clinician diagnostic performance with machine learning-based decision support systems: a systematic review. JAMA Netw Open 2021;4:e211276. https://doi.org/10.1001/jamanetworkopen.2021.1276. sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Translating research techniques into clinical practice, establishing accountability for clinical decisions, and determining the extent of disclosure of machine learning integration are major challenges associated with the use of machine learning models.

“Overall, translating research techniques into clinical practice, establishing accountability for clinical decisions [129], and the extent of disclosure of ML integration are the major challenges associated with the use of ML models [135].”

• cites [129] Peiffer-Smadja, N, Rawson, TM, Ahmad, R, Buchard, A, Georgiou, P, Lescure, FX, et al.. Machine learning for clinical decision support in infectious diseases: a narrative review of current applications. Clin Microbiol Infect 2020;26:584–95. https://doi.org/10.1016/j.cmi.2019.09.009. sourceNot checked — no full text of the source
• cites [135] Kelly, CJ, Karthikesalingam, A, Suleyman, M, Corrado, G, King, D. Key challenges for delivering clinical impact with artificial intelligence. BMC Med 2019;17:195. https://doi.org/10.1186/s12916-019-1426-2. sourcePartly supported

  Split into 2 atoms — 1 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 1 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Regulatory classification of software depends on its intended purpose: if primarily associated with in vitro diagnostic data, it falls under the IVDR; otherwise, it falls under the MDR.

“The regulatory classification depends on the software's intended purpose; if primarily associated with In Vitro Diagnostic (IVD) data, it falls under IVDR [132, 134], otherwise, it falls under MDR [133].”

• cites [132] The European Parliament and the Council of the European Union. European Parliament and Council Regulation (EU) 2017/746 of 5 April 2017 on in vitro diagnostic medical devices. Available at: https://eur-lex.europa.eu/eli/reg/2017/746/oj [Accessed 16 Nov 2023].Not checked yet

  Not checked in the free run.

• cites [134] Vanstapel, F, Orth, M, Streichert, T, Capoluongo, ED, Oosterhuis, WP, Çubukçu, HC, et al.. ISO 15189 is a sufficient instrument to guarantee high-quality manufacture of laboratory developed tests for in-house-use conform requirements of the European In-Vitro-Diagnostics Regulation. Clin Chem Lab Med 2023;61:608–26. https://doi.org/10.1515/cclm-2023-0045. sourceSupported
• cites [133] The European Parliament and the Council of the European Union. European Parliament and Council Regulation (EU) 2017/745 of 5 April 2017 on medical devices. Available at: https://eur-lex.europa.eu/eli/reg/2017/745/oj [Accessed 16 Nov 2023].Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

There are currently no standardized evaluation techniques for explainable AI tools.

“Furthermore, there are no standardized evaluation techniques for XAI tools currently available [122].”

• cites [122] Markus, AF, Kors, JA, Rijnbeek, PR. The role of explainability in creating trustworthy artificial intelligence for health care: a comprehensive survey of the terminology, design choices, and evaluation strategies. J Biomed Inf 2021;113:103655. https://doi.org/10.1016/j.jbi.2020.103655. sourceSupported
  Experimental — accuracy not yet measured

May combine two claims.

Farrell and colleagues' artificial neural network model achieved 92.1% accuracy and an AUC of 0.977 for identifying incorrectly labeled samples.

“Notably, the ANN model yielded impressive results, achieving an accuracy of 92.1 % and an area under the curve (AUC) of 0.977.”

• cites [24] Farrell, CJ. Identifying mislabelled samples: machine learning models exceed human performance. Ann Clin Biochem 2021;58:650–2. https://doi.org/10.1177/00045632211032991. sourceNot checked — no full text of the source

May combine two claims.

Farrell and colleagues' artificial neural network model achieved 90.6% sensitivity, 94.5% specificity, and 92.5% accuracy for detecting wrong blood in tube errors.

“Through the use of an ANN model, they achieved noteworthy sensitivity of 90.6 %, specificity of 94.5 %, and accuracy of 92.5 %.”

• cites [25] Farrell, CL. Decision support or autonomous artificial intelligence? The case of wrong blood in tube errors. Clin Chem Lab Med 2022;60:1993–7. https://doi.org/10.1515/cclm-2021-0873. sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Fang and colleagues studied whether clotted specimens could be identified by analyzing coagulation test results.

“In a study conducted by Fang et al., the identification of clotted specimens was explored through the analysis of coagulation test results [23].”

• cites [23] Fang, K, Dong, Z, Chen, X, Zhu, J, Zhang, B, You, J, et al.. Using machine learning to identify clotted specimens in coagulation testing. Clin Chem Lab Med 2021;59:1289–97. https://doi.org/10.1515/cclm-2021-0081. sourceNot checked — no full text of the source

May combine two claims.

Zhou and colleagues' deep belief network-based delta check method outperformed other approaches, with an AUC of 0.977, accuracy of 93.1%, true positive rate of 92.9%, and true negative rate of 93.3%.

“Notably, the DBN-based delta check method outperformed other approaches, yielding an impressive AUC of 0.977, accuracy of 93.1 %, true positive rate (TPR) of 92.9 %, and true negative rate (TNR) of 93.3 %.”

• cites [26] Zhou, R, Liang, YF, Cheng, HL, Wang, W, Huang, DW, Wang, Z, et al.. A highly accurate delta check method using deep learning for detection of sample mix-up in the clinical laboratory. Clin Chem Lab Med 2022;60:1984–92. https://doi.org/10.1515/cclm-2021-1171. sourceNot checked — no full text of the source

May combine two claims.

Fang and colleagues' backpropagation neural network model reached an AUC of 0.971, accuracy of 0.953, specificity of 0.967, and sensitivity of 0.940 for identifying clotted samples.

“Utilizing standard and momentum backpropagation neural networks (BPNNs), the researchers developed a high-performance model, achieving an impressive area under the curve (AUC) of 0.971, accuracy of 0.953, specificity of 0.967, and sensitivity of 0.940.”

• cites [23] Fang, K, Dong, Z, Chen, X, Zhu, J, Zhang, B, You, J, et al.. Using machine learning to identify clotted specimens in coagulation testing. Clin Chem Lab Med 2021;59:1289–97. https://doi.org/10.1515/cclm-2021-0081. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Feature importance is one of the most used global explanation methods.

“As seen in Table 1, feature importance is one of the most used global explanation methods.”

Probably not a substantial claim: may not pass S4 (based).

Machine-learning clinical decision support is changing laboratory medicine by letting laboratories handle more work with less direct human oversight across the pre-analytical, analytical, and post-analytical stages.

“They are revolutionizing laboratory medicine, enabling labs to work more efficiently with less human supervision across pre-analytical, analytical, and post-analytical phases.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Putting machine-learning tools into practice brings difficulties such as uncertain models, opaque algorithms, and loss of professional skills.

“Despite contributions of the ML tools at all analytical phases, their integration presents challenges like potential model uncertainties, blackbox algorithms, and deskilling of professionals.”

May combine two claims.

Automated machine-learning tools were created to make model building easier and allow people without data-science expertise to produce models.

“Recently, due to the complexity inherent in model development, automated ML tools were also introduced to streamline the process, enabling non-experts to create models.”

• Also stated in: § Automated machine learning · plain

  “To overcome this problem, automated machine learning (AutoML) tools have been introduced which can build high-quality machine learning models for specific tasks without human expertise.”

• cites [10] Waring, J, Lindvall, C, Umeton, R. Automated machine learning: review of the state-of-the-art and opportunities for healthcare. Artif Intell Med 2020;104:101822. https://doi.org/10.1016/j.artmed.2020.101822. sourceNot checked — no full text of the source
• cites [11] Topcu, D, Bayraktar, N. Searching for the urine osmolality surrogate: an automated machine learning approach. Clin Chem Lab Med 2022;60:1911–20. https://doi.org/10.1515/cclm-2022-0415. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

AI and machine-learning technologies offer valuable opportunities for thorough analysis and interpretation of large healthcare datasets.

“In light of this burgeoning data landscape, the utilization of AI and ML technologies undoubtedly offers valuable prospects for the comprehensive analysis and interpretation of this vast wealth of information [22].”

• cites [22] Gruson, D, Helleputte, T, Rousseau, P, Gruson, D. Data science, artificial intelligence, and machine learning: opportunities for laboratory medicine and the value of positive regulation. Clin Biochem 2019;69:1–7. https://doi.org/10.1016/j.clinbiochem.2019.04.013. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Healthcare practitioners face growing amounts and types of data from imaging, genomics, proteomics, clinical observations, and personal and environmental records.

“Moreover, in the present era, healthcare practitioners are confronted with an escalating volume and diversity of data, encompassing various domains such as imaging, genomics, proteomics, clinical observations, as well as personal and environmental records.”

May combine two claims.

Continuous monitoring and real-time data analysis allow early detection of health problems and timely interventions, improving care and lowering healthcare costs.

“Patient monitoring: Continuous monitoring and realtime data analysis enable early detection of health issues and timely interventions, improving patient care and reducing healthcare costs [18].”

• cites [18] Hurvitz, N, Azmanov, H, Kesler, A, Ilan, Y. Establishing a second-generation artificial intelligence-based system for improving diagnosis, treatment, and monitoring of patients with rare diseases. Eur J Hum Genet 2021;29:1485–90. https://doi.org/10.1038/s41431-021-00928-4. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Using AI and machine learning can markedly improve the accuracy and efficiency of clinical decision support systems, helping clinicians make better-informed and more effective decisions.

“By harnessing the power of AI/ML, the accuracy and efficiency of CDS systems can be significantly improved, allowing healthcare professionals to make more informed and effective decisions in their clinical practice.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

AI, especially its machine-learning subset, has great potential to transform healthcare, including clinical laboratories.

“The implications of AI, particularly its subset known as machine learning (ML), hold immense potential for revolutionizing the healthcare sector including clinical laboratories.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Machine-learning models that use laboratory results have drawn growing attention in the scientific literature.

“The implementation of ML models incorporating laboratory results has attracted increasing attention in the scientific literature.”

Probably not a substantial claim: may not pass S3 (load-bearing).

AI has led to complex devices that include software using AI technologies.

“While the invention of microprocessors triggered total laboratory automation, nowadays, AI has paved the way for complex devices that include software with AI technologies.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The coming together of cyber-physical systems, IoT, cloud computing, machine learning, and AI is both a current and expected reality that is transforming laboratory medicine.

“Consequently, the convergence of cyber-physical systems, IoT, cloud computing, ML, and AI presents a tangible reality in the present and an anticipated future reality, revolutionizing laboratory medicine [19,20].”

• cites [19] Ialongo, C, Bernardini, S. Total laboratory automation has the potential to be the field of application of artificial intelligence: the cyber-physical system and “Automation 4.0”. Clin Chem Lab Med 2019;57:e279–81. https://doi.org/10.1515/cclm-2019-0226. sourceNot checked — no full text of the source
• cites [20] Neumaier, M. Diagnostics 4.0: the medical laboratory in digital health. Clin Chem Lab Med 2019;57:343–8. https://doi.org/10.1515/cclm-2018-1088. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Laboratory medicine is undergoing major changes because of automation and AI.

“The field of laboratory medicine is undergoing significant transformations as a result of two prominent technological advancements: automation and AI [14].”

• cites [14] Naugler, C, Church, DL. Automation and artificial intelligence in the clinical laboratory. Crit Rev Clin Lab Sci 2019;56:98–110. https://doi.org/10.1080/10408363.2018.1561640. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

AI and machine-learning models have considerable potential to increase the role of in vitro diagnostics in clinical decisions, thereby improving patient care and outcomes.

“AI/ML models hold substantial potential in improving the contribution of IVD to clinical decision-making, thereby enhancing patient care and outcomes.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Machine-learning models have been used in the pre-analytical, analytical, and post-analytical stages of the total testing process.

“These ML models have been applied across various stages of the total testing process, including preanalytical, analytical, and post-analytical phases, as given in Figure 2.”

• AssumptionRests on an assumption: The articles selected from reputable laboratory medicine journals are representative of machine-learning applications across the total testing process.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Adding external validation supports unbiased testing and helps ensure a model works across different demographic groups.

“Overall, the integration of external validation enables unbiased testing, ensuring the model's applicability across different demographics.”

Probably not a substantial claim: may not pass S3 (load-bearing).

External validation serves as a strong way to judge a model's external validity and its applicability to a wide range of patients.

“It serves as a robust mechanism to assess the model's external validity and determine its applicability across a broad range of patients or individuals.”

Probably not a substantial claim: may not pass S4 (based).

Machine-learning clinical decision support in healthcare can substantially improve clinical decisions.

“In conclusion, ML-based CDSS in healthcare can greatly enhance clinical decision-making.”

• AssumptionRests on an assumption: The articles selected from reputable laboratory medicine journals are representative of machine-learning applications across the total testing process.

Classified as probably not a substantial claim.

The review chose AI and machine-learning articles from reputable laboratory medicine journals that showed usefulness in the pre-analytical, analytical, and post-analytical phases.

“Specifically, we selected articles on AI/ML from reputable laboratory medicine journals that have demonstrated utility in the preanalytical, analytical, and post-analytical phases within total testing process.”

• AssumptionRests on an assumption: The articles selected from reputable laboratory medicine journals are representative of machine-learning applications across the total testing process.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Automating parts of machine-learning development that need expertise lets healthcare professionals build, validate, and deploy solutions faster, which can improve patient care quality.

“Moreover, by automating some of the ML development components requiring expertise, healthcare professionals can more rapidly build, validate, and deploy ML solutions, and therefore more readily improve the quality of healthcare for patients.”

May combine two claims.

AI and machine-learning models can forecast disease progression and patient outcomes, helping providers make informed care and intervention decisions.

“Prognostics: Predicting disease progression and patient outcomes is another critical application [14].These models can help healthcare providers make informed decisions about patient care and intervention.”

• cites [14] Naugler, C, Church, DL. Automation and artificial intelligence in the clinical laboratory. Crit Rev Clin Lab Sci 2019;56:98–110. https://doi.org/10.1080/10408363.2018.1561640. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Recent progress in AI has shown promise for improving clinical decision support systems.

“However, recent advancements in AI have demonstrated promising potential in enhancing CDS systems [13].”

• cites [13] Baron, JM, Kurant, DE, Dighe, AS. Machine learning and other emerging decision support tools. Clin Lab Med 2019;39:319–31. https://doi.org/10.1016/j.cll.2019.01.010. sourceNot checked — no full text of the source

May combine two claims.

AutoML tools automate feature engineering, model development, hyperparameter optimization, and performance evaluation.

“These tools aim to automate the pipeline of ML model development, including feature engineering, model development, hyperparameter optimization, and performance evaluation, all of which typically necessitate experienced users as shown in Figure 1 [10,11].”

• cites [10] Waring, J, Lindvall, C, Umeton, R. Automated machine learning: review of the state-of-the-art and opportunities for healthcare. Artif Intell Med 2020;104:101822. https://doi.org/10.1016/j.artmed.2020.101822. sourceNot checked — no full text of the source
• cites [11] Topcu, D, Bayraktar, N. Searching for the urine osmolality surrogate: an automated machine learning approach. Clin Chem Lab Med 2022;60:1911–20. https://doi.org/10.1515/cclm-2022-0415. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Clinical laboratories are shifting through the adoption of sophisticated automated systems powered by AI software and advanced robotics.

“Clinical laboratories are experiencing a paradigm shift through the integration of sophisticated automated systems empowered by AI-driven software and advanced robotic technologies.”

Classified as probably not a substantial claim.

The review examined the chosen articles using criteria that included data features, machine-learning methods, programming languages and packages, best model performance, explainability, reported limitations, and merits or outcomes.

“Furthermore, to ensure the quality and relevance of our selected articles, we meticulously scrutinized them based on the following criteria: data features, ML methods employed, programming languages and packages utilized, the performance of the best ML model, considerations of model explainability, reported study limitations, and merits or outcomes observed.”

May combine two claims.

AutoML lowers the demand for data scientists and lets domain experts build machine-learning applications with little statistical or machine-learning training.

“Therefore, they reduce the need for data scientists, enabling domain experts to create ML applications with minimal requirements of statistical and ML expertise [11].”

• cites [11] Topcu, D, Bayraktar, N. Searching for the urine osmolality surrogate: an automated machine learning approach. Clin Chem Lab Med 2022;60:1911–20. https://doi.org/10.1515/cclm-2022-0415. sourceNot checked — no full text of the source

May combine two claims.

Clinical decision support systems help healthcare professionals interpret test results, reduce subjective interpretation, and lessen inconsistencies.

“Clinical decision support (CDS) systems play a crucial role in assisting healthcare professionals in the interpretation of test results, mitigating interpretative subjectivity, and minimizing inconsistencies [12].”

• cites [12] Bright, TJ, Wong, A, Dhurjati, R, Bristow, E, Bastian, L, Coeytaux, RR, et al.. Effect of clinical decision-support systems: a systematic review. Ann Intern Med 2012;157:29–43. https://doi.org/10.7326/0003-4819-157-1-201207030-00450. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

Obtaining varied datasets is difficult, and the intricacy of models can restrict their clinical usefulness.

“Additionally, acquiring diverse datasets is hard, and models' complexity can limit clinical use.”

Probably not a substantial claim: may not pass S3 (load-bearing).

This convergence allows laboratories to complete larger amounts of work with less need for extensive human involvement.

“This convergence enables the accomplishment of greater volumes of work with a reduced need for extensive human intervention within the laboratory setting [19].”

• cites [19] Ialongo, C, Bernardini, S. Total laboratory automation has the potential to be the field of application of artificial intelligence: the cyber-physical system and “Automation 4.0”. Clin Chem Lab Med 2019;57:e279–81. https://doi.org/10.1515/cclm-2019-0226. sourceNot checked — no full text of the source

May combine two claims.

AI and machine-learning models can support early and accurate disease diagnosis by analyzing patient data and test results, lowering misdiagnosis risk and improving outcomes.

“Diagnosis: AI/ML models can aid in the early and accurate diagnosis of diseases by analyzing patient data and test results [14], reducing the risk of misdiagnosis and improving patient outcomes.”

• cites [14] Naugler, C, Church, DL. Automation and artificial intelligence in the clinical laboratory. Crit Rev Clin Lab Sci 2019;56:98–110. https://doi.org/10.1080/10408363.2018.1561640. sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Clinical decision support powered by AI and machine learning can help clinicians create personalized treatment plans that account for patient-specific factors and current medical evidence.

“Treatment planning: CDS systems powered by AI/ML can assist healthcare professionals in creating personalized treatment plans [16], considering patient-specific factors and the latest medical research.”

• cites [16] Wen, Z, Wang, S, Yang, DM, Xie, Y, Chen, M, Bishop, J, et al.. Deep learning in digital pathology for personalized treatment plans of cancer patients. Semin Diagn Pathol 2023;40:109–19. https://doi.org/10.1053/j.semdp.2023.02.003. sourceNot checked — no full text of the source

May combine two claims.

In the study by Yang 2020, the GBDT model achieved an AUC of 0.838, sensitivity of 0.758, and specificity of 0.740 on an independent dataset.

“The GBDT model achieved an AUC of ., sensitivity of ., and specificity of . on an independent dataset.”

• cites Yang 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Demirci 2016, the ANN model achieved a sensitivity of 92.2% and a specificity of 99.6%.

“Sensitivity of . % and specificity of . %.”

• cites Demirci 2016Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Wu 2022, the random forest model achieved a sensitivity of 100% and a specificity of 88.9%.

“Random forest achieved a sensitivity of  % and specificity of . %.”

• cites Wu 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Rashidi 2021, the logistic regression model achieved an accuracy of 96%, sensitivity of 92.3%, specificity of 97.7%, and AUC of 0.96.

“The logistic regression model achieved an accuracy of  %, a sensitivity of . %, a specificity of . %, and an AUC of ..”

• cites Rashidi 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Feng 2022, the random forest model achieved an AUC of 0.948, specificity of 0.967, accuracy of 0.915, PPV of 0.942, and NPV of 0.901 on the external validation dataset.

“The random forest model achieved an AUC of ., specificity of ., accuracy of ., PPV of ., and NPV of . on the external validation dataset.”

• cites Feng 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims. The quote only partly states this claim.

In the study by He 2021, the random forest model achieved an 85.0% detection rate for Down syndrome while ensuring a false positive rate of 5%.

“The model achieved an . % detection rate for DS while ensuring a false positive rate of  %.”

• cites He 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Ng 2021, the random forest model classified B-cell malignancies with a sensitivity of 83.57%, specificity of 99.26%, PPV of 96.69%, NPV of 95.87%, and accuracy of 96.02%.

“The model classified B-cell malignancies with a sensitivity of . %, specificity of . %, PPV of . % and NPV of . %, and an accuracy of . %.”

• cites Ng 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Hauser 2021, the XGBoost/LASSO models achieved AUC values of 0.59-0.67 for 2-5 years, 0.75-0.80 for 0.5-1 year, and 0.87-0.92 at diagnosis.

“AUC values: - years (.-.),.- year (.-.),at diagnosis (.-.).”

• cites Hauser 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Lafuente-Ganuza 2020, the decision tree/random forest model achieved an NPV of 100% and a PPV of 87% for early-onset pre-eclampsia.

“The NPV for earlyonset pre-eclampsia is  %, while the PPV is  %.”

• cites Lafuente-Ganuza 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Mitani and colleagues' study was limited to a simulation-based approach and lacked external validation.

“Nonetheless, it is worth mentioning that this study was limited to a simulation-based approach and lacked external validation [27]”

• cites [27] Mitani, T, Doi, S, Yokota, S, Imai, T, Ohe, K. Highly accurate and explainable detection of specimen mix-up using a machine learning model. Clin Chem Lab Med 2020;58:375–83. https://doi.org/10.1515/cclm-2019-0534. sourceNot checked — no full text of the source

May combine two claims.

In the study by Cadamuro 2023, ChatGPT ratings on a 1-6 scale were 5-6 for relevance, 4-5 for correctness, 3-4 for helpfulness, and 5-6 for safety.

“Regarding rating from  (very low) to  (very high): relevance -, correctness -, helpfulness -, safety -.”

• cites Cadamuro 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Yang 2022, the SVM model achieved an accuracy of 100% and a kappa score of 1.000.

“The SVM model achieved an accuracy of  % and a kappa score of ..”

• cites Yang 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Lin 2022, the deep learning model achieved an AUC of 0.8948 for hypercalcemia and 0.7723 for hypocalcemia.

“The area under the curve (AUC) for hypercalcemia was ., while for hypocalcemia it was ..”

• cites Lin 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Cabitza 2020, KNN and RF models achieved AUCs between 0.75 and 0.78 and external validation specificities between 0.92 and 0.96.

“KNN and RF models achieved AUCs of .-.and external validation specificities of .-..”

• cites Cabitza 2020Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Ng 2015, the SVM model achieved an AUC of 0.96, accuracy of 0.95, sensitivity of 1, and specificity of 0.91.

“The SVM model achieved an AUC of ., an accuracy of ., a sensitivity of , and a specificity of ..”

• cites Ng 2015Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Topcu 2022, the AutoML models achieved R2 values between 0.70 and 0.83, and around 70-84% of results were within the agreed limit.

“The R value ranged between . and ., and around - % of the results were within the agreed limit.”

• cites Topcu 2022 sourceNot checked — no full text of the source

May combine two claims. The quote only partly states this claim.

In the study by Simonson 2021, the EnsembleCNN classifier achieved an accuracy of 88.0%, precision of 82.4%, recall of 67.7%, F1 score of 74.3%, and AUC of 0.92.

“The EnsembleCNN classifier achieved an accuracy of . %, precision of . %, recall (sensitivity) of . %, and F score of . %, with an AUC of ..”

• cites Simonson 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Han 2021, the SVM model achieved an AUC of 0.777 for identifying benign breast lesions and an AUC of 0.824 for identifying malignant breast lesions.

“The SVM model achieved an AUC of . for identifying benign breast lesions and an AUC of . for identifying malignant breast lesions.”

• cites Han 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Simonson 2022, the CNN model achieved an accuracy of 94%, AUROC of 89%, recall of 78%, F1 score of 0.62, and precision of 51%.

“Accuracy  %, AUROC  %, recall  %, F score ., precision  %.”

• cites Simonson 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

The quote only partly states this claim.

In the study by Soerensen 2022, the ANN model achieved an AUC of 0.79.

“ANN achieved an AUC of ..”

• cites Soerensen 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims. The quote only partly states this claim.

In the study by Gui 2023, the SVM model achieved a sensitivity of 93.1%, a specificity of 100%, and an AUC of 0.966.

“SVM achieved a sensitivity of . %, a specificity of  %, and an AUC of ..”

• cites Gui 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Lee 2022, the random forest model achieved a sensitivity of 93.7%, specificity of 84.4%, and accuracy of 88.0%.

“Sensitivity . %, specificity . %, accuracy . %.”

• cites Lee 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Wilkes 2018, the weighted-subspace RF model achieved an AUC of 0.955 for abnormal vs. normal classification and an AUC of 0.873 for multiclass classification.

“Weighted-subspace RF achieved discrimination performance with an AUC of . for abnormal vs. normal classification, and an AUC of . for multiclass classification.”

• cites Wilkes 2018Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Hu 2023, the CNN model achieved an average accuracy of 99.82%, sensitivity of 93.17%, and specificity of 99.93%.

“The average accuracy of . %, sensitivity of . %, and specificity of . %.”

• cites Hu 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Peña-Bautista 2019, the ANN model achieved an accuracy of 0.882, sensitivity of 88.2%, and specificity of 76.9%.

“ANN achieved an accuracy of ., with a sensitivity of . % and a specificity of . %.”

• cites Peña-Bautista 2019Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims. The quote only partly states this claim.

In the study by Constantinescu 2022, the ML-based steroidomics models assigned primary aldosteronism probabilities ranging from 80% to 100% (median 90%) in PA patients and from 2% to 90% (median 21%) in non-PA patients.

“Primary aldosteronism (PA) probabilities ranged from  to  % (median  %) in PA patients, and from  to  % (median  %) in non-PA patients.”

• cites Constantinescu 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

In the study by Hu 2021, the multilayer perceptron neural network achieved an AUC of 0.942.

“Multilayer perceptron neural network: AUC ..”

• cites Hu 2021Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

In the study by Streun 2022, the random forest model correctly classified 88% of the test set, including 80% of positive samples and 96% of negative samples.

“The model correctly classified  % of the test set, with  % of positive samples and  % of negative samples.”

• cites Streun 2022Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

The quote only partly states this claim.

In the study by Luo 2016, the random forest regression model achieved an AUC of 0.97 in discriminating normal and abnormal ferritin results.

“High accuracy (AUC: .) in discriminating normal and abnormal ferritin results.”

• cites Luo 2016Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims. The quote only partly states this claim.

A random forest model integrating conventional quality control rules, EWMA, and CUSUM charts achieved a false rejection probability of 0.0048 and the highest error detection rate for errors less than one standard deviation.

“The RF model achieved a low false rejection probability of 0.0048 and demonstrated the highest error detection rate for errors less than one standard deviation.”

• cites ÇubukçuNot matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [40] Çubukçu, HC. Performance evaluation of internal quality control rules, EWMA, CUSUM, and the novel machine learning model. Turk J Biochem 2021;46:661–70. https://doi.org/10.1515/tjb-2021-0199. sourceNot checked — no full text of the source

May combine two claims.

The random forest quality control model achieved an accuracy of 75%, a sensitivity of 71.3%, a specificity of 99.6%, and a false positive rate of 0.45%.

“The model exhibited an accuracy of 75 %, sensitivity of 71.3 %, specificity of 99.6 %, and a low false positive rate (FPR) of 0.45 %.”

• cites [39] Zhou, R, Wang, W, Padoan, A, Wang, Z, Feng, X, Han, Z, et al.. Traceable machine learning real-time quality control based on patient data. Clin Chem Lab Med 2022;60:1998–2004. https://doi.org/10.1515/cclm-2022-0548. sourceNot checked — no full text of the source

May combine two claims.

A CNN model for assessing serum quality achieved AUC values of 0.989 for hemolysis, 0.996 for icterus, and 0.993 for lipemia.

“The CNN model demonstrated exceptional performance, achieving high area under the curve (AUC) values for detecting hemolysis (0.989), icterus (0.996), and lipemia (0.993).”

• cites Yang et al.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [31] Yang, C, Li, D, Sun, D, Zhang, S, Zhang, P, Xiong, Y, et al.. A deep learning-based system for assessment of serum quality using sample images. Clin Chim Acta 2022;531:254–60. https://doi.org/10.1016/j.cca.2022.04.010. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Machine learning and image analysis methods show versatility and potential in cell analysis.

“These studies utilize various techniques, including statistical feature extraction, dimensionality reduction, and deep learning models, demonstrating the versatility and potential of ML and image analysis methods in the field of cell analysis.”

May combine two claims.

An MLP-ANN model reduced wasted tests for κ-free light chains and λ-free light chains by 69.4% and 70.8%, respectively.

“The implementation of MLP-ANN resulted in a remarkable reduction in wasted tests for κ-FLC and λ-FLC, with reductions of 69.4 and 70.8 %, respectively.”

• cites Ialongo (2017)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [29] Ialongo, C, Pieri, M, Bernardini, S. Smart management of sample dilution using an artificial neural network to achieve streamlined processes and saving resources: the automated nephelometric testing of serum free light chain as case study. Clin Chem Lab Med 2017;55:231–6. https://doi.org/10.1515/cclm-2016-0263. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Explainability in machine learning models is essential for reliable AI in clinical contexts, regulatory compliance, and determining responsibility for AI faults.

“For reliable AI in clinical contexts, regulatory compliance, and determining who is responsible for AI faults, explainability in ML models is essential [111].”

• cites [111] Carobene, A, Cabitza, F, Bernardini, S, Gopalan, R, Lennerz, JK, Weir, C, et al.. Where is laboratory medicine headed in the next decade? Partnership model for efficient integration and adoption of artificial intelligence into medical laboratories. Clin Chem Lab Med 2023;61:535–43. https://doi.org/10.1515/cclm-2023-0352. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

A CNN model for erythrocyte morphology classification achieved a recall of 92.70%, a precision of 89.39%, and a correct classification frequency of 90.60%.

“Utilizing a dataset of 3,737 labeled cells, the CNN model achieved a recall of 92.70 %, a precision of 89.39 %, and a correct classification frequency of 90.60 %.”

• cites Durant et al.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [35] Durant, TJS, Olson, EM, Schulz, WL, Torres, R. Very deep convolutional neural networks for morphologic classification of erythrocytes. Clin Chem 2017;63:1847–55. https://doi.org/10.1373/clinchem.2017.276345. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Some pre-analytical machine learning studies have limitations including disparities in sample distribution, lack of external validation, and limited explainability.

“However, some studies have limitations such as disparities in sample distribution, lack of external validation, and limited explainability.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Black box models, such as artificial neural networks, lack explainability, although methods exist to enhance interpretability.

“Black box models, like artificial neural networks, lack explainability, although methods exist to enhance interpretability [125].”

• cites [125] Cabitza, F, Rasoini, R, Gensini, GF. Unintended consequences of machine learning in medicine. JAMA 2017;318:517–8. https://doi.org/10.1001/jama.2017.7797. sourceNot checked — no full text of the source

May combine two claims.

A CNN model for detecting fetal nucleated red blood cells achieved an accuracy of 98.5%, a sensitivity of 96.5%, and a specificity of 100%.

“The CNN model achieved an accuracy of 98.5 %, sensitivity of 96.5 %, and specificity of 100 % for fNRBC detection.”

• cites Sun et al.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [37] Sun, C, Wang, R, Zhao, L, Han, L, Ma, S, Liang, D, et al.. A computer-aided diagnosis system of fetal nucleated red blood cells with convolutional neural network. Arch Pathol Lab Med 2022;146:1395–401. https://doi.org/10.5858/arpa.2021-0142-oa. sourceNot checked — no full text of the source

May combine two claims.

Machine learning models demonstrated superior capabilities in preventing wrong blood in tube errors, reducing potential harm to patients.

“The application of ML models demonstrated superior capabilities in preventing WBIT errors, effectively reducing potential harm to patients [28].”

• cites [28] Rosenbaum, MW, Baron, JM. Using machine learning-based multianalyte delta checks to detect wrong blood in tube errors. Am J Clin Pathol 2018;150:555–66. https://doi.org/10.1093/ajcp/aqy085. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

In-house devices are permissible only when no equivalent medical device is available on the market, as specified by the MDR and IVDR.

“Additionally, the utilization of in-house devices is permissible only when no equivalent medical device is available in the market, as specified by MDR and IVDR [132,133].”

• cites [132] The European Parliament and the Council of the European Union. European Parliament and Council Regulation (EU) 2017/746 of 5 April 2017 on in vitro diagnostic medical devices. Available at: https://eur-lex.europa.eu/eli/reg/2017/746/oj [Accessed 16 Nov 2023].Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [133] The European Parliament and the Council of the European Union. European Parliament and Council Regulation (EU) 2017/745 of 5 April 2017 on medical devices. Available at: https://eur-lex.europa.eu/eli/reg/2017/745/oj [Accessed 16 Nov 2023].Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

The IFCC working group published practical recommendations to facilitate the implementation of machine learning in laboratory medicine.

“The IFCC working group has contributed significantly by publishing practical recommendations aimed at facilitating the implementation of ML in laboratory medicine [126].”

• cites [126] Master, SR, Badrick, TC, Bietenbeck, A, Haymond, S. Machine learning in laboratory medicine: recommendations of the IFCC Working Group. Clin Chem 2023;69:690–8. https://doi.org/10.1093/clinchem/hvad055. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S4 (based).

Machine learning-based clinical decision support systems in laboratory medicine and healthcare improve clinical decision-making across precision medicine to population health.

“The use of machine learning (ML)-based clinical decision support systems (CDSS) in laboratory medicine and healthcare, from precision medicine to population health, greatly improves clinical decision-making.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

AI/ML-based clinical decision support systems are currently being implemented in clinical practice as Software as a Medical Device.

“AI/ML-based clinical decision support (CDS) systems are currently being implemented in clinical practice as Software as a Medical Device (SaMD).”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The adoption of machine learning is impacted by infrastructure constraints, data quality issues, and privacy protection.

“The adoption of ML is impacted by infrastructure constraints, data quality issues, and privacy protection [129].”

• cites [129] Peiffer-Smadja, N, Rawson, TM, Ahmad, R, Buchard, A, Georgiou, P, Lescure, FX, et al.. Machine learning for clinical decision support in infectious diseases: a narrative review of current applications. Clin Microbiol Infect 2020;26:584–95. https://doi.org/10.1016/j.cmi.2019.09.009. sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Principles from the FDA, Health Canada, and the MHRA guide medical device development and provide a framework for good machine-learning practices.

“Additionally, the principles presented by the Food and Drug Administration (FDA), Health Canada, and the Medicines & Healthcare Products Regulatory Agency serve to guide the development of medical devices, offering a comprehensive framework for the adoption of “good machine-learning practices” [144].”

• cites [144] The U.S. Food and Drug Administration, Health Canada, and the United Kingdom’s Medicines and Healthcare products Regulatory Agency. Good machine learning practice for medical device development: guiding principles, October 2021. Available at: https://www.fda.gov/medical-devices/software-medical-device-samd/good-machine-learning-practice-medical-device-development-guiding-principles [Accessed 1 May 2023].Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

Guiding documents collectively advance and standardize machine learning practices in laboratory medicine and medical device development, improving outcomes and quality and safety in healthcare.

“These guiding documents collectively contribute to the advancement and standardization of ML practices in laboratory medicine and medical device development, fostering improved outcomes and ensuring the highest level of quality and safety in the healthcare industry.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Clearer definitions and compliance standards for AI-based Software as a Medical Device have been established in Europe through the Medical Device Regulation and the In Vitro Medical Device Regulation.

“Although ISO 15189:2022 does not explicitly delineate AI-based SaMD [131], clearer definitions and compliance standards have been established in Europe through the Medical Device Regulation (MDR) and In Vitro Medical Device Regulation (IVDR) [132, 133].”

• cites [132] The European Parliament and the Council of the European Union. European Parliament and Council Regulation (EU) 2017/746 of 5 April 2017 on in vitro diagnostic medical devices. Available at: https://eur-lex.europa.eu/eli/reg/2017/746/oj [Accessed 16 Nov 2023].Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites [133] The European Parliament and the Council of the European Union. European Parliament and Council Regulation (EU) 2017/745 of 5 April 2017 on medical devices. Available at: https://eur-lex.europa.eu/eli/reg/2017/745/oj [Accessed 16 Nov 2023].Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

ISO 15189:2022 does not explicitly delineate AI-based Software as a Medical Device.

“Although ISO 15189:2022 does not explicitly delineate AI-based SaMD [131], clearer definitions and compliance standards have been established in Europe through the Medical Device Regulation (MDR) and In Vitro Medical Device Regulation (IVDR) [132, 133].”

• cites [131] International Organization for Standardization. ISO 15189:2012 Medical laboratories – requirements for quality and competence; 2022.Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Liability and responsibility for AI and machine learning-assisted clinical decision-making remain uncertain.

“Other issues to be concerned about include the stilluncertain liability and responsibility for AI and ML-assisted clinical decision-making.”

• cites [13] Baron, JM, Kurant, DE, Dighe, AS. Machine learning and other emerging decision support tools. Clin Lab Med 2019;39:319–31. https://doi.org/10.1016/j.cll.2019.01.010. sourceNot checked — no full text of the source
• cites [129] Peiffer-Smadja, N, Rawson, TM, Ahmad, R, Buchard, A, Georgiou, P, Lescure, FX, et al.. Machine learning for clinical decision support in infectious diseases: a narrative review of current applications. Clin Microbiol Infect 2020;26:584–95. https://doi.org/10.1016/j.cmi.2019.09.009. sourceNot checked — no full text of the source