Regulatory oversight, causal inference, and safe and effective health care machine learning

Open-Source Clinical Machine Learning Models: Critical Appraisal of Feasibility, Advantages, and Challenges

Machine learning applications promise to augment clinical capabilities and at least 64 models have already been approved by the US Food and Drug Administration. These tools are developed, shared, and used in an environment in which regulations and market forces remain immature. An important consideration when evaluating this environment is the introduction of open-source solutions in which innovations are freely shared; such solutions have long been a facet of digital culture. We discuss the feasibility and implications of open-source machine learning in a health care infrastructure built upon proprietary information. The decreased cost of development as compared to drugs and devices, a longstanding culture of open-source products in other industries, and the beginnings of machine learning–friendly regulatory pathways together allow for the development and deployment of open-source machine learning models. Such tools have distinct advantages including enhanced product integrity, customizability, and lower cost, leading to increased access. However, significant questions regarding engineering concerns about implementation infrastructure and model safety, a lack of incentives from intellectual property protection, and nebulous liability rules significantly complicate the ability to develop such open-source models. Ultimately, the reconciliation of open-source machine learning and the proprietary information–driven health care environment requires that policymakers, regulators, and health care organizations actively craft a conducive market in which innovative developers will continue to both work and collaborate.

Digital Biomarkers in Psychiatric Research: Data Protection Qualifications in a Complex Ecosystem

Psychiatric research traditionally relies on subjective observation, which is time-consuming and labor-intensive. The widespread use of digital devices, such as smartphones and wearables, enables the collection and use of vast amounts of user-generated data as "digital biomarkers." These tools may also support increased participation of psychiatric patients in research and, as a result, the production of research results that are meaningful to them. However, sharing mental health data and research results may expose patients to discrimination and stigma risks, thus discouraging participation. To earn and maintain participants' trust, the first essential requirement is to implement an appropriate data governance system with a clear and transparent allocation of data protection duties and responsibilities among the actors involved in the process. These include sponsors, investigators, operators of digital tools, as well as healthcare service providers and biobanks/databanks. While previous works have proposed practical solutions to this end, there is a lack of consideration of positive data protection law issues in the extant literature. To start filling this gap, this paper discusses the GDPR legal qualifications of controller, processor, and joint controllers in the complex ecosystem unfolded by the integration of digital biomarkers in psychiatric research, considering their implications and proposing some general practical recommendations.

Generalization in Clinical Prediction Models: The Blessing and Curse of Measurement Indicator Variables

Supplemental Digital Content is available in the text.

OBJECTIVE:

Specific factors affecting generalizability of clinical prediction models are poorly understood. Our main objective was to investigate how measurement indicator variables affect external validity in clinical prediction models for predicting onset of vasopressor therapy.

DESIGN:

We fit logistic regressions on retrospective cohorts to predict vasopressor onset using two classes of variables: seemingly objective clinical variables (vital signs and laboratory measurements) and more subjective variables denoting recency of measurements.

SETTING:

Three cohorts from two tertiary-care academic hospitals in geographically distinct regions, spanning general inpatient and critical care settings.

PATIENTS:

Each cohort consisted of adult patients (age greater than or equal to 18 yr at time of hospitalization), with lengths of stay between 6 and 600 hours, and who did not receive vasopressors in the first 6 hours of hospitalization or ICU admission. Models were developed on each of the three derivation cohorts and validated internally on the derivation cohort and externally on the other two cohorts.

INTERVENTIONS:

None.

MEASUREMENTS AND MAIN RESULTS:

The prevalence of vasopressors was 0.9% in the general inpatient cohort and 12.4% and 11.5% in the two critical care cohorts. Models utilizing both classes of variables performed the best in-sample, with C-statistics for predicting vasopressor onset in 4 hours of 0.862 (95% CI, 0.844–0.879), 0.822 (95% CI, 0.793–0.852), and 0.889 (95% CI, 0.880–0.898). Models solely using the subjective variables denoting measurement recency had poor external validity. However, these practice-driven variables helped adjust for differences between the two hospitals and led to more generalizable models using clinical variables.

CONCLUSIONS:

We developed and externally validated models for predicting the onset of vasopressors. We found that practice-specific features denoting measurement recency improved local performance and also led to more generalizable models if they are adjusted for during model development but discarded at validation. The role of practice-specific features such as measurement indicators in clinical prediction modeling should be carefully considered if the goal is to develop generalizable models.

Rise of Clinical Studies in the Field of Machine Learning: A Review of Data Registered in ClinicalTrials.gov

Although advances in machine-learning healthcare applications promise great potential for innovative medical care, few data are available on the translational status of these new technologies. We aimed to provide a comprehensive characterization of the development and status quo of clinical studies in the field of machine learning. For this purpose, we performed a registry-based analysis of machine-learning-related studies that were published and first available in the ClinicalTrials.gov database until 2020, using the database's study classification. In total, n = 358 eligible studies could be included in the analysis. Of these, 82% were initiated by academic institutions/university (hospitals) and 18% by industry sponsors. A total of 96% were national and 4% international. About half of the studies (47%) had at least one recruiting location in a country in North America, followed by Europe (37%) and Asia (15%). Most of the studies reported were initiated in the medical field of imaging (12%), followed by cardiology, psychiatry, anesthesia/intensive care medicine (all 11%) and neurology (10%). Although the majority of the clinical studies were still initiated in an academic research context, the first industry-financed projects on machine-learning-based algorithms are becoming visible. The number of clinical studies with machine-learning-related applications and the variety of medical challenges addressed serve to indicate their increasing importance in future clinical care. Finally, they also set a time frame for the adjustment of medical device-related regulation and governance.