Ethical and regulatory challenges of large language models in medicine

With the rapid growth of interest in and use of large language models (LLMs) across various industries, we are facing some crucial and profound ethical concerns, especially in the medical field. The unique technical architecture and purported emergent abilities of LLMs differentiate them substantially from other artificial intelligence (AI) models and natural language processing techniques used, necessitating a nuanced understanding of LLM ethics. In this Viewpoint, we highlight ethical concerns stemming from the perspectives of users, developers, and regulators, notably focusing on data privacy and rights of use, data provenance, intellectual property contamination, and broad applications and plasticity of LLMs. A comprehensive framework and mitigating strategies will be imperative for the responsible integration of LLMs into medical practice, ensuring alignment with ethical principles and safeguarding against potential societal risks.

Scalable Approach to Consumer Wearable Postmarket Surveillance: Development and Validation Study

Background

With the capability to render prediagnoses, consumer wearables have the potential to affect subsequent diagnoses and the level of care in the health care delivery setting. Despite this, postmarket surveillance of consumer wearables has been hindered by the lack of codified terms in electronic health records (EHRs) to capture wearable use.

Objective

We sought to develop a weak supervision-based approach to demonstrate the feasibility and efficacy of EHR-based postmarket surveillance on consumer wearables that render atrial fibrillation (AF) prediagnoses.

Methods

We applied data programming, where labeling heuristics are expressed as code-based labeling functions, to detect incidents of AF prediagnoses. A labeler model was then derived from the predictions of the labeling functions using the Snorkel framework. The labeler model was applied to clinical notes to probabilistically label them, and the labeled notes were then used as a training set to fine-tune a classifier called Clinical-Longformer. The resulting classifier identified patients with an AF prediagnosis. A retrospective cohort study was conducted, where the baseline characteristics and subsequent care patterns of patients identified by the classifier were compared against those who did not receive a prediagnosis.

Results

The labeler model derived from the labeling functions showed high accuracy (0.92; F1-score=0.77) on the training set. The classifier trained on the probabilistically labeled notes accurately identified patients with an AF prediagnosis (0.95; F1-score=0.83). The cohort study conducted using the constructed system carried enough statistical power to verify the key findings of the Apple Heart Study, which enrolled a much larger number of participants, where patients who received a prediagnosis tended to be older, male, and White with higher CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes, stroke, vascular disease, age 65-74 years, sex category) scores (P<.001). We also made a novel discovery that patients with a prediagnosis were more likely to use anticoagulants (525/1037, 50.63% vs 5936/16,560, 35.85%) and have an eventual AF diagnosis (305/1037, 29.41% vs 262/16,560, 1.58%). At the index diagnosis, the existence of a prediagnosis did not distinguish patients based on clinical characteristics, but did correlate with anticoagulant prescription (P=.004 for apixaban and P=.01 for rivaroxaban).

Conclusions

Our work establishes the feasibility and efficacy of an EHR-based surveillance system for consumer wearables that render AF prediagnoses. Further work is necessary to generalize these findings for patient populations at other sites.

Ensuring useful adoption of generative artificial intelligence in healthcare

OBJECTIVES

This article aims to examine how generative artificial intelligence (AI) can be adopted with the most value in health systems, in response to the Executive Order on AI.

MATERIALS AND METHODS

We reviewed how technology has historically been deployed in healthcare, and evaluated recent examples of deployments of both traditional AI and generative AI (GenAI) with a lens on value.

RESULTS

Traditional AI and GenAI are different technologies in terms of their capability and modes of current deployment, which have implications on value in health systems.

DISCUSSION

Traditional AI when applied with a framework top-down can realize value in healthcare. GenAI in the short term when applied top-down has unclear value, but encouraging more bottom-up adoption has the potential to provide more benefit to health systems and patients.

CONCLUSION

GenAI in healthcare can provide the most value for patients when health systems adapt culturally to grow with this new technology and its adoption patterns.

A Nationwide Network of Health AI Assurance Laboratories

Importance

Given the importance of rigorous development and evaluation standards needed of artificial intelligence (AI) models used in health care, nationwide accepted procedures to provide assurance that the use of AI is fair, appropriate, valid, effective, and safe are urgently needed.

Observations

While there are several efforts to develop standards and best practices to evaluate AI, there is a gap between having such guidance and the application of such guidance to both existing and new AI models being developed. As of now, there is no publicly available, nationwide mechanism that enables objective evaluation and ongoing assessment of the consequences of using health AI models in clinical care settings.

Conclusion and Relevance

The need to create a public-private partnership to support a nationwide health AI assurance labs network is outlined here. In this network, community best practices could be applied for testing health AI models to produce reports on their performance that can be widely shared for managing the lifecycle of AI models over time and across populations and sites where these models are deployed.

President Biden's Executive Order on Artificial Intelligence-Implications for Health Care Organizations

This Viewpoint discusses a recent executive order by US President Joe Biden about the development and implementation of AI, including the role of government vs the private sector and how the order may affect health care.

Lessons Learned from a Multi-Site, Team-Based Serious Illness Care Program Implementation at an Academic Medical Center

Background: Patients with serious illness benefit from conversations to share prognosis and explore goals and values. To address this, we implemented Ariadne Labs' Serious Illness Care Program (SICP) at Stanford Health Care. Objective: Improve quantity, timing, and quality of serious illness conversations. Methods: Initial implementation followed Ariadne Labs' SICP framework. We later incorporated a team-based approach that included nonphysician care team members. Outcomes included number of patients with documented conversations according to clinician role and practice location. Machine learning algorithms were used in some settings to identify eligible patients. Results: Ambulatory oncology and hospital medicine were our largest implementation sites, engaging 4707 and 642 unique patients in conversations, respectively. Clinicians across eight disciplines engaged in these conversations. Identified barriers that included leadership engagement, complex workflows, and patient identification. Conclusion: Several factors contributed to successful SICP implementation across clinical sites: innovative clinical workflows, machine learning based predictive algorithms, and nonphysician care team member engagement.

The Stanford Medicine data science ecosystem for clinical and translational research

Objective

To describe the infrastructure, tools, and services developed at Stanford Medicine to maintain its data science ecosystem and research patient data repository for clinical and translational research.

Materials and Methods

The data science ecosystem, dubbed the Stanford Data Science Resources (SDSR), includes infrastructure and tools to create, search, retrieve, and analyze patient data, as well as services for data deidentification, linkage, and processing to extract high-value information from healthcare IT systems. Data are made available via self-service and concierge access, on HIPAA compliant secure computing infrastructure supported by in-depth user training.

Results

The Stanford Medicine Research Data Repository (STARR) functions as the SDSR data integration point, and includes electronic medical records, clinical images, text, bedside monitoring data and HL7 messages. SDSR tools include tools for electronic phenotyping, cohort building, and a search engine for patient timelines. The SDSR supports patient data collection, reproducible research, and teaching using healthcare data, and facilitates industry collaborations and large-scale observational studies.

Discussion

Research patient data repositories and their underlying data science infrastructure are essential to realizing a learning health system and advancing the mission of academic medical centers. Challenges to maintaining the SDSR include ensuring sufficient financial support while providing researchers and clinicians with maximal access to data and digital infrastructure, balancing tool development with user training, and supporting the diverse needs of users.

Conclusion

Our experience maintaining the SDSR offers a case study for academic medical centers developing data science and research informatics infrastructure.

Ranitidine Use and Incident Cancer in a Multinational Cohort

Importance

Ranitidine, the most widely used histamine-2 receptor antagonist (H2RA), was withdrawn because of N-nitrosodimethylamine impurity in 2020. Given the worldwide exposure to this drug, the potential risk of cancer development associated with the intake of known carcinogens is an important epidemiological concern.

Objective

To examine the comparative risk of cancer associated with the use of ranitidine vs other H2RAs.

Design, Setting, and Participants

This new-user active comparator international network cohort study was conducted using 3 health claims and 9 electronic health record databases from the US, the United Kingdom, Germany, Spain, France, South Korea, and Taiwan. Large-scale propensity score (PS) matching was used to minimize confounding of the observed covariates with negative control outcomes. Empirical calibration was performed to account for unobserved confounding. All databases were mapped to a common data model. Database-specific estimates were combined using random-effects meta-analysis. Participants included individuals aged at least 20 years with no history of cancer who used H2RAs for more than 30 days from January 1986 to December 2020, with a 1-year washout period. Data were analyzed from April to September 2021.

Exposure

The main exposure was use of ranitidine vs other H2RAs (famotidine, lafutidine, nizatidine, and roxatidine).

Main Outcomes and Measures

The primary outcome was incidence of any cancer, except nonmelanoma skin cancer. Secondary outcomes included all cancer except thyroid cancer, 16 cancer subtypes, and all-cause mortality.

Results

Among 1 183 999 individuals in 11 databases, 909 168 individuals (mean age, 56.1 years; 507 316 [55.8%] women) were identified as new users of ranitidine, and 274 831 individuals (mean age, 58.0 years; 145 935 [53.1%] women) were identified as new users of other H2RAs. Crude incidence rates of cancer were 14.30 events per 1000 person-years (PYs) in ranitidine users and 15.03 events per 1000 PYs among other H2RA users. After PS matching, cancer risk was similar in ranitidine compared with other H2RA users (incidence, 15.92 events per 1000 PYs vs 15.65 events per 1000 PYs; calibrated meta-analytic hazard ratio, 1.04; 95% CI, 0.97-1.12). No significant associations were found between ranitidine use and any secondary outcomes after calibration.

Conclusions and Relevance

In this cohort study, ranitidine use was not associated with an increased risk of cancer compared with the use of other H2RAs. Further research is needed on the long-term association of ranitidine with cancer development.

Creation and Adoption of Large Language Models in Medicine

Importance

There is increased interest in and potential benefits from using large language models (LLMs) in medicine. However, by simply wondering how the LLMs and the applications powered by them will reshape medicine instead of getting actively involved, the agency in shaping how these tools can be used in medicine is lost.

Observations

Applications powered by LLMs are increasingly used to perform medical tasks without the underlying language model being trained on medical records and without verifying their purported benefit in performing those tasks.

Conclusions and Relevance

The creation and use of LLMs in medicine need to be actively shaped by provisioning relevant training data, specifying the desired benefits, and evaluating the benefits via testing in real-world deployments.

DEPLOYR: a technical framework for deploying custom real-time machine learning models into the electronic medical record

OBJECTIVE

Heatlhcare institutions are establishing frameworks to govern and promote the implementation of accurate, actionable, and reliable machine learning models that integrate with clinical workflow. Such governance frameworks require an accompanying technical framework to deploy models in a resource efficient, safe and high-quality manner. Here we present DEPLOYR, a technical framework for enabling real-time deployment and monitoring of researcher-created models into a widely used electronic medical record system.

MATERIALS AND METHODS

We discuss core functionality and design decisions, including mechanisms to trigger inference based on actions within electronic medical record software, modules that collect real-time data to make inferences, mechanisms that close-the-loop by displaying inferences back to end-users within their workflow, monitoring modules that track performance of deployed models over time, silent deployment capabilities, and mechanisms to prospectively evaluate a deployed model's impact.

RESULTS

We demonstrate the use of DEPLOYR by silently deploying and prospectively evaluating 12 machine learning models trained using electronic medical record data that predict laboratory diagnostic results, triggered by clinician button-clicks in Stanford Health Care's electronic medical record.

DISCUSSION

Our study highlights the need and feasibility for such silent deployment, because prospectively measured performance varies from retrospective estimates. When possible, we recommend using prospectively estimated performance measures during silent trials to make final go decisions for model deployment.

CONCLUSION

Machine learning applications in healthcare are extensively researched, but successful translations to the bedside are rare. By describing DEPLOYR, we aim to inform machine learning deployment best practices and help bridge the model implementation gap.

Use of Electronic Health Record Data for Drug Safety Signal Identification: A Scoping Review

INTRODUCTION

Pharmacovigilance programs protect patient health and safety by identifying adverse event signals through postmarketing surveillance of claims data and spontaneous reports. Electronic health records (EHRs) provide new opportunities to address limitations of traditional approaches and promote discovery-oriented pharmacovigilance.

METHODS

To evaluate the current state of EHR-based medication safety signal identification, we conducted a scoping literature review of studies aimed at identifying safety signals from routinely collected patient-level EHR data. We extracted information on study design, EHR data elements utilized, analytic methods employed, drugs and outcomes evaluated, and key statistical and data analysis choices.

RESULTS

We identified 81 eligible studies. Disproportionality methods were the predominant analytic approach, followed by data mining and regression. Variability in study design makes direct comparisons difficult. Studies varied widely in terms of data, confounding adjustment, and statistical considerations.

CONCLUSION

Despite broad interest in utilizing EHRs for safety signal identification, current efforts fail to leverage the full breadth and depth of available data or to rigorously control for confounding. The development of best practices and application of common data models would promote the expansion of EHR-based pharmacovigilance.

The shaky foundations of large language models and foundation models for electronic health records

The success of foundation models such as ChatGPT and AlphaFold has spurred significant interest in building similar models for electronic medical records (EMRs) to improve patient care and hospital operations. However, recent hype has obscured critical gaps in our understanding of these models' capabilities. In this narrative review, we examine 84 foundation models trained on non-imaging EMR data (i.e., clinical text and/or structured data) and create a taxonomy delineating their architectures, training data, and potential use cases. We find that most models are trained on small, narrowly-scoped clinical datasets (e.g., MIMIC-III) or broad, public biomedical corpora (e.g., PubMed) and are evaluated on tasks that do not provide meaningful insights on their usefulness to health systems. Considering these findings, we propose an improved evaluation framework for measuring the benefits of clinical foundation models that is more closely grounded to metrics that matter in healthcare.

Characterizing subgroup performance of probabilistic phenotype algorithms within older adults: a case study for dementia, mild cognitive impairment, and Alzheimer's and Parkinson's diseases

Objective

Biases within probabilistic electronic phenotyping algorithms are largely unexplored. In this work, we characterize differences in subgroup performance of phenotyping algorithms for Alzheimer's disease and related dementias (ADRD) in older adults.

Materials and methods

We created an experimental framework to characterize the performance of probabilistic phenotyping algorithms under different racial distributions allowing us to identify which algorithms may have differential performance, by how much, and under what conditions. We relied on rule-based phenotype definitions as reference to evaluate probabilistic phenotype algorithms created using the Automated PHenotype Routine for Observational Definition, Identification, Training and Evaluation framework.

Results

We demonstrate that some algorithms have performance variations anywhere from 3% to 30% for different populations, even when not using race as an input variable. We show that while performance differences in subgroups are not present for all phenotypes, they do affect some phenotypes and groups more disproportionately than others.

Discussion

Our analysis establishes the need for a robust evaluation framework for subgroup differences. The underlying patient populations for the algorithms showing subgroup performance differences have great variance between model features when compared with the phenotypes with little to no differences.

Conclusion

We have created a framework to identify systematic differences in the performance of probabilistic phenotyping algorithms specifically in the context of ADRD as a use case. Differences in subgroup performance of probabilistic phenotyping algorithms are not widespread nor do they occur consistently. This highlights the great need for careful ongoing monitoring to evaluate, measure, and try to mitigate such differences.

Predicting patients who are likely to develop Lupus Nephritis of those newly diagnosed with Systemic Lupus Erythematosus

Patients diagnosed with systemic lupus erythematosus (SLE) suffer from a decreased quality of life, an increased risk of medical complications, and an increased risk of death. In particular, approximately 50% of SLE patients progress to develop lupus nephritis, which oftentimes leads to life-threatening end stage renal disease (ESRD) and requires dialysis or kidney transplant1. The challenge is that lupus nephritis is diagnosed via a kidney biopsy, which is typically performed only after noticeable decreased kidney function, leaving little room for proactive or preventative measures. The ability to predict which patients are most likely to develop lupus nephritis has the potential to shift lupus nephritis disease management from reactive to proactive. We present a clinically useful prediction model to predict which patients with newly diagnosed SLE will go on to develop lupus nephritis in the next five years.

Clinical utility gains from incorporating comorbidity and geographic location information into risk estimation equations for atherosclerotic cardiovascular disease

OBJECTIVE

There are over 363 customized risk models of the American College of Cardiology and the American Heart Association (ACC/AHA) pooled cohort equations (PCE) in the literature, but their gains in clinical utility are rarely evaluated. We build new risk models for patients with specific comorbidities and geographic locations and evaluate whether performance improvements translate to gains in clinical utility.

MATERIALS AND METHODS

We retrain a baseline PCE using the ACC/AHA PCE variables and revise it to incorporate subject-level information of geographic location and 2 comorbidity conditions. We apply fixed effects, random effects, and extreme gradient boosting (XGB) models to handle the correlation and heterogeneity induced by locations. Models are trained using 2 464 522 claims records from Optum©'s Clinformatics® Data Mart and validated in the hold-out set (N = 1 056 224). We evaluate models' performance overall and across subgroups defined by the presence or absence of chronic kidney disease (CKD) or rheumatoid arthritis (RA) and geographic locations. We evaluate models' expected utility using net benefit and models' statistical properties using several discrimination and calibration metrics.

RESULTS

The revised fixed effects and XGB models yielded improved discrimination, compared to baseline PCE, overall and in all comorbidity subgroups. XGB improved calibration for the subgroups with CKD or RA. However, the gains in net benefit are negligible, especially under low exchange rates.

CONCLUSIONS

Common approaches to revising risk calculators incorporating extra information or applying flexible models may enhance statistical performance; however, such improvement does not necessarily translate to higher clinical utility. Thus, we recommend future works to quantify the consequences of using risk calculators to guide clinical decisions.

Assessing the net benefit of machine learning models in the presence of resource constraints

OBJECTIVE

The objective of this study is to provide a method to calculate model performance measures in the presence of resource constraints, with a focus on net benefit (NB).

MATERIALS AND METHODS

To quantify a model's clinical utility, the Equator Network's TRIPOD guidelines recommend the calculation of the NB, which reflects whether the benefits conferred by intervening on true positives outweigh the harms conferred by intervening on false positives. We refer to the NB achievable in the presence of resource constraints as the realized net benefit (RNB), and provide formulae for calculating the RNB.

RESULTS

Using 4 case studies, we demonstrate the degree to which an absolute constraint (eg, only 3 available intensive care unit [ICU] beds) diminishes the RNB of a hypothetical ICU admission model. We show how the introduction of a relative constraint (eg, surgical beds that can be converted to ICU beds for very high-risk patients) allows us to recoup some of the RNB but with a higher penalty for false positives.

DISCUSSION

RNB can be calculated in silico before the model's output is used to guide care. Accounting for the constraint changes the optimal strategy for ICU bed allocation.

CONCLUSIONS

This study provides a method to account for resource constraints when planning model-based interventions, either to avoid implementations where constraints are expected to play a larger role or to design more creative solutions (eg, converted ICU beds) to overcome absolute constraints when possible.

Use of Machine Learning and Lay Care Coaches to Increase Advance Care Planning Conversations for Patients With Metastatic Cancer

PURPOSE

Patients with metastatic cancer benefit from advance care planning (ACP) conversations. We aimed to improve ACP using a computer model to select high-risk patients, with shorter predicted survival, for conversations with providers and lay care coaches. Outcomes included ACP documentation frequency and end-of-life quality measures.

METHODS

In this study of a quality improvement initiative, providers in four medical oncology clinics received Serious Illness Care Program training. Two clinics (thoracic/genitourinary) participated in an intervention, and two (cutaneous/sarcoma) served as controls. ACP conversations were documented in a centralized form in the electronic medical record. In the intervention, providers and care coaches received weekly e-mails highlighting upcoming clinic patients with < 2 year computer-predicted survival and no prior prognosis documentation. Care coaches contacted these patients for an ACP conversation (excluding prognosis). Providers were asked to discuss and document prognosis.

RESULTS

In the four clinics, 4,968 clinic visits by 1,251 patients met inclusion criteria (metastatic cancer with no prognosis previously documented). In their first visit, 28% of patients were high-risk (< 2 year predicted survival). Preintervention, 3% of both intervention and control clinic patients had ACP documentation during a visit. By intervention end (February 2021), 35% of intervention clinic patients had ACP documentation compared with 3% of control clinic patients. Providers' prognosis documentation rate also increased in intervention clinics after the intervention (2%-27% in intervention clinics, P < .0001; 0%-1% in control clinics). End-of-life care intensity was similar in intervention versus control clinics, but patients with ≥ 1 provider ACP edit met fewer high-intensity care measures (P = .04).

CONCLUSION

Combining a computer prognosis model with care coaches increased ACP documentation.

Investigating real-world consequences of biases in commonly used clinical calculators

OBJECTIVES

To evaluate whether one summary metric of calculator performance sufficiently conveys equity across different demographic subgroups, as well as to evaluate how calculator predictive performance affects downstream health outcomes.

STUDY DESIGN

We evaluate 3 commonly used clinical calculators-Model for End-Stage Liver Disease (MELD), CHA2DS2-VASc, and simplified Pulmonary Embolism Severity Index (sPESI)-on the cohort extracted from the Stanford Medicine Research Data Repository, following the cohort selection process as described in respective calculator derivation papers.

METHODS

We quantified the predictive performance of the 3 clinical calculators across sex and race. Then, using the clinical guidelines that guide care based on these calculators' output, we quantified potential disparities in subsequent health outcomes.

RESULTS

Across the examined subgroups, the MELD calculator exhibited worse performance for female and White populations, CHA2DS2-VASc calculator for the male population, and sPESI for the Black population. The extent to which such performance differences translated into differential health outcomes depended on the distribution of the calculators' scores around the thresholds used to trigger a care action via the corresponding guidelines. In particular, under the old guideline for CHA2DS2-VASc, among those who would not have been offered anticoagulant therapy, the Hispanic subgroup exhibited the highest rate of stroke.

CONCLUSIONS

Clinical calculators, even when they do not include variables such as sex and race as inputs, can have very different care consequences across those subgroups. These differences in health care outcomes across subgroups can be explained by examining the distribution of scores and their calibration around the thresholds encoded in the accompanying care guidelines.

A computational approach to measure the linguistic characteristics of psychotherapy timing, responsiveness, and consistency

Although individual psychotherapy is generally effective for a range of mental health conditions, little is known about the moment-to-moment language use of effective therapists. Increased access to computational power, coupled with a rise in computer-mediated communication (telehealth), makes feasible the large-scale analyses of language use during psychotherapy. Transparent methodological approaches are lacking, however. Here we present novel methods to increase the efficiency of efforts to examine language use in psychotherapy. We evaluate three important aspects of therapist language use - timing, responsiveness, and consistency - across five clinically relevant language domains: pronouns, time orientation, emotional polarity, therapist tactics, and paralinguistic style. We find therapist language is dynamic within sessions, responds to patient language, and relates to patient symptom diagnosis but not symptom severity. Our results demonstrate that analyzing therapist language at scale is feasible and may help answer longstanding questions about specific behaviors of effective therapists.

Considerations in the reliability and fairness audits of predictive models for advance care planning

Multiple reporting guidelines for artificial intelligence (AI) models in healthcare recommend that models be audited for reliability and fairness. However, there is a gap of operational guidance for performing reliability and fairness audits in practice. Following guideline recommendations, we conducted a reliability audit of two models based on model performance and calibration as well as a fairness audit based on summary statistics, subgroup performance and subgroup calibration. We assessed the Epic End-of-Life (EOL) Index model and an internally developed Stanford Hospital Medicine (HM) Advance Care Planning (ACP) model in 3 practice settings: Primary Care, Inpatient Oncology and Hospital Medicine, using clinicians' answers to the surprise question (“Would you be surprised if [patient X] passed away in [Y years]?”) as a surrogate outcome. For performance, the models had positive predictive value (PPV) at or above 0.76 in all settings. In Hospital Medicine and Inpatient Oncology, the Stanford HM ACP model had higher sensitivity (0.69, 0.89 respectively) than the EOL model (0.20, 0.27), and better calibration (O/E 1.5, 1.7) than the EOL model (O/E 2.5, 3.0). The Epic EOL model flagged fewer patients (11%, 21% respectively) than the Stanford HM ACP model (38%, 75%). There were no differences in performance and calibration by sex. Both models had lower sensitivity in Hispanic/Latino male patients with Race listed as “Other.” 10 clinicians were surveyed after a presentation summarizing the audit. 10/10 reported that summary statistics, overall performance, and subgroup performance would affect their decision to use the model to guide care; 9/10 said the same for overall and subgroup calibration. The most commonly identified barriers for routinely conducting such reliability and fairness audits were poor demographic data quality and lack of data access. This audit required 115 person-hours across 8–10 months. Our recommendations for performing reliability and fairness audits include verifying data validity, analyzing model performance on intersectional subgroups, and collecting clinician-patient linkages as necessary for label generation by clinicians. Those responsible for AI models should require such audits before model deployment and mediate between model auditors and impacted stakeholders.

Performance and usability testing of an automated tool for detection of peripheral artery disease using electronic health records

Peripheral artery disease (PAD) is a common cardiovascular disorder that is frequently underdiagnosed, which can lead to poorer outcomes due to lower rates of medical optimization. We aimed to develop an automated tool to identify undiagnosed PAD and evaluate physician acceptance of a dashboard representation of risk assessment. Data were derived from electronic health records (EHR). We developed and compared traditional risk score models to novel machine learning models. For usability testing, primary and specialty care physicians were recruited and interviewed until thematic saturation. Data from 3168 patients with PAD and 16,863 controls were utilized. Results showed a deep learning model that utilized time engineered features outperformed random forest and traditional logistic regression models (average AUCs 0.96, 0.91 and 0.81, respectively), P < 0.0001. Of interviewed physicians, 75% were receptive to an EHR-based automated PAD model. Feedback emphasized workflow optimization, including integrating risk assessments directly into the EHR, using dashboard designs that minimize clicks, and providing risk assessments for clinically complex patients. In conclusion, we demonstrate that EHR-based machine learning models can accurately detect risk of PAD and that physicians are receptive to automated risk detection for PAD. Future research aims to prospectively validate model performance and impact on patient outcomes.

Assessment of Adherence to Reporting Guidelines by Commonly Used Clinical Prediction Models From a Single Vendor: A Systematic Review

IMPORTANCE

Various model reporting guidelines have been proposed to ensure clinical prediction models are reliable and fair. However, no consensus exists about which model details are essential to report, and commonalities and differences among reporting guidelines have not been characterized. Furthermore, how well documentation of deployed models adheres to these guidelines has not been studied.

OBJECTIVES

To assess information requested by model reporting guidelines and whether the documentation for commonly used machine learning models developed by a single vendor provides the information requested.

EVIDENCE REVIEW

MEDLINE was queried using machine learning model card and reporting machine learning from November 4 to December 6, 2020. References were reviewed to find additional publications, and publications without specific reporting recommendations were excluded. Similar elements requested for reporting were merged into representative items. Four independent reviewers and 1 adjudicator assessed how often documentation for the most commonly used models developed by a single vendor reported the items.

FINDINGS

From 15 model reporting guidelines, 220 unique items were identified that represented the collective reporting requirements. Although 12 items were commonly requested (requested by 10 or more guidelines), 77 items were requested by just 1 guideline. Documentation for 12 commonly used models from a single vendor reported a median of 39% (IQR, 37%-43%; range, 31%-47%) of items from the collective reporting requirements. Many of the commonly requested items had 100% reporting rates, including items concerning outcome definition, area under the receiver operating characteristics curve, internal validation, and intended clinical use. Several items reported half the time or less related to reliability, such as external validation, uncertainty measures, and strategy for handling missing data. Other frequently unreported items related to fairness (summary statistics and subgroup analyses, including for race and ethnicity or sex).

CONCLUSIONS AND RELEVANCE

These findings suggest that consistent reporting recommendations for clinical predictive models are needed for model developers to share necessary information for model deployment. The many published guidelines would, collectively, require reporting more than 200 items. Model documentation from 1 vendor reported the most commonly requested items from model reporting guidelines. However, areas for improvement were identified in reporting items related to model reliability and fairness. This analysis led to feedback to the vendor, which motivated updates to the documentation for future users.

Nursing Workflow Change in a COVID-19 Inpatient Unit Following the Deployment of Inpatient Telehealth: Observational Study Using a Real-Time Locating System

BACKGROUND

The COVID-19 pandemic prompted widespread implementation of telehealth, including in the inpatient setting, with the goals to reduce potential pathogen exposure events and personal protective equipment (PPE) utilization. Nursing workflow adaptations in these novel environments are of particular interest given the association between nursing time at the bedside and patient safety. Understanding the frequency and duration of nurse-patient encounters following the introduction of a novel telehealth platform in the context of COVID-19 may therefore provide insight into downstream impacts on patient safety, pathogen exposure, and PPE utilization.

OBJECTIVE

The aim of this study was to evaluate changes in nursing workflow relative to prepandemic levels using a real-time locating system (RTLS) following the deployment of inpatient telehealth on a COVID-19 unit.

METHODS

In March 2020, telehealth was installed in patient rooms in a COVID-19 unit and on movable carts in 3 comparison units. The existing RTLS captured nurse movement during 1 pre- and 5 postpandemic stages (January-December 2020). Change in direct nurse-patient encounters, time spent in patient rooms per encounter, and total time spent with patients per shift relative to baseline were calculated. Generalized linear models assessed difference-in-differences in outcomes between COVID-19 and comparison units. Telehealth adoption was captured and reported at the unit level.

RESULTS

Change in frequency of encounters and time spent per encounter from baseline differed between the COVID-19 and comparison units at all stages of the pandemic (all P<.001). Frequency of encounters decreased (difference-in-differences range -6.6 to -14.1 encounters) and duration of encounters increased (difference-in-differences range 1.8 to 6.2 minutes) from baseline to a greater extent in the COVID-19 units relative to the comparison units. At most stages of the pandemic, the change in total time nurses spent in patient rooms per patient per shift from baseline did not differ between the COVID-19 and comparison units (all P>.17). The primary COVID-19 unit quickly adopted telehealth technology during the observation period, initiating 15,088 encounters that averaged 6.6 minutes (SD 13.6) each.

CONCLUSIONS

RTLS movement data suggest that total nursing time at the bedside remained unchanged following the deployment of inpatient telehealth in a COVID-19 unit. Compared to other units with shared mobile telehealth units, the frequency of nurse-patient in-person encounters decreased and the duration lengthened on a COVID-19 unit with in-room telehealth availability, indicating "batched" redistribution of work to maintain total time at bedside relative to prepandemic periods. The simultaneous adoption of telehealth suggests that virtual care was a complement to, rather than a replacement for, in-person care. However, study limitations preclude our ability to draw a causal link between nursing workflow change and telehealth adoption. Thus, further evaluation is needed to determine potential downstream implications on disease transmission, PPE utilization, and patient safety.

Monitoring Approaches for a Pediatric Chronic Kidney Disease Machine Learning Model

OBJECTIVE

The purpose of this study is to evaluate the ability of three metrics to monitor for a reduction in performance of a chronic kidney disease (CKD) model deployed at a pediatric hospital.

METHODS

The CKD risk model estimates a patient's risk of developing CKD 3 to 12 months following an inpatient admission. The model was developed on a retrospective dataset of 4,879 admissions from 2014 to 2018, then run silently on 1,270 admissions from April to October, 2019. Three metrics were used to monitor its performance during the silent phase: (1) standardized mean differences (SMDs); (2) performance of a "membership model"; and (3) response distribution analysis. Observed patient outcomes for the 1,270 admissions were used to calculate prospective model performance and the ability of the three metrics to detect performance changes.

RESULTS

The deployed model had an area under the receiver-operator curve (AUROC) of 0.63 in the prospective evaluation, which was a significant decrease from an AUROC of 0.76 on retrospective data (p = 0.033). Among the three metrics, SMDs were significantly different for 66/75 (88%) of the model's input variables (p <0.05) between retrospective and deployment data. The membership model was able to discriminate between the two settings (AUROC = 0.71, p <0.0001) and the response distributions were significantly different (p <0.0001) for the two settings.

CONCLUSION

This study suggests that the three metrics examined could provide early indication of performance deterioration in deployed models' performance.

DLMM as a lossless one-shot algorithm for collaborative multi-site distributed linear mixed models

Linear mixed models are commonly used in healthcare-based association analyses for analyzing multi-site data with heterogeneous site-specific random effects. Due to regulations for protecting patients’ privacy, sensitive individual patient data (IPD) typically cannot be shared across sites. We propose an algorithm for fitting distributed linear mixed models (DLMMs) without sharing IPD across sites. This algorithm achieves results identical to those achieved using pooled IPD from multiple sites (i.e., the same effect size and standard error estimates), hence demonstrating the lossless property. The algorithm requires each site to contribute minimal aggregated data in only one round of communication. We demonstrate the lossless property of the proposed DLMM algorithm by investigating the associations between demographic and clinical characteristics and length of hospital stay in COVID-19 patients using administrative claims from the UnitedHealth Group Clinical Discovery Database. We extend this association study by incorporating 120,609 COVID-19 patients from 11 collaborative data sources worldwide.

A lossless, one-shot and privacy-preserving distributed algorithm was revealed for fitting linear mixed models on multi-site data. The algorithm was applied to a study of 120,609 COVID-19 patients using only minimal aggregated data from each of 14 sites.

Building a Learning Health System: Creating an Analytical Workflow for Evidence Generation to Inform Institutional Clinical Care Guidelines

Background  One key aspect of a learning health system (LHS) is utilizing data generated during care delivery to inform clinical care. However, institutional guidelines that utilize observational data are rare and require months to create, making current processes impractical for more urgent scenarios such as those posed by the COVID-19 pandemic. There exists a need to rapidly analyze institutional data to drive guideline creation where evidence from randomized control trials are unavailable.

Objectives  This article provides a background on the current state of observational data generation in institutional guideline creation and details our institution's experience in creating a novel workflow to (1) demonstrate the value of such a workflow, (2) demonstrate a real-world example, and (3) discuss difficulties encountered and future directions.

Methods  Utilizing a multidisciplinary team of database specialists, clinicians, and informaticists, we created a workflow for identifying and translating a clinical need into a queryable format in our clinical data warehouse, creating data summaries and feeding this information back into clinical guideline creation.

Results  Clinical questions posed by the hospital medicine division were answered in a rapid time frame and informed creation of institutional guidelines for the care of patients with COVID-19. The cost of setting up a workflow, answering the questions, and producing data summaries required around 300 hours of effort and $300,000 USD.

Conclusion  A key component of an LHS is the ability to learn from data generated during care delivery. There are rare examples in the literature and we demonstrate one such example along with proposed thoughts of ideal multidisciplinary team formation and deployment.

Predictors of diagnostic transition from major depressive disorder to bipolar disorder: a retrospective observational network study

Many patients with bipolar disorder (BD) are initially misdiagnosed with major depressive disorder (MDD) and are treated with antidepressants, whose potential iatrogenic effects are widely discussed. It is unknown whether MDD is a comorbidity of BD or its earlier stage, and no consensus exists on individual conversion predictors, delaying BD's timely recognition and treatment. We aimed to build a predictive model of MDD to BD conversion and to validate it across a multi-national network of patient databases using the standardization afforded by the Observational Medical Outcomes Partnership (OMOP) common data model. Five "training" US databases were retrospectively analyzed: IBM MarketScan CCAE, MDCR, MDCD, Optum EHR, and Optum Claims. Cyclops regularized logistic regression models were developed on one-year MDD-BD conversion with all standard covariates from the HADES PatientLevelPrediction package. Time-to-conversion Kaplan-Meier analysis was performed up to a decade after MDD, stratified by model-estimated risk. External validation of the final prediction model was performed across 9 patient record databases within the Observational Health Data Sciences and Informatics (OHDSI) network internationally. The model's area under the curve (AUC) varied 0.633-0.745 (µ = 0.689) across the five US training databases. Nine variables predicted one-year MDD-BD transition. Factors that increased risk were: younger age, severe depression, psychosis, anxiety, substance misuse, self-harm thoughts/actions, and prior mental disorder. AUCs of the validation datasets ranged 0.570-0.785 (µ = 0.664). An assessment algorithm was built for MDD to BD conversion that allows distinguishing as much as 100-fold risk differences among patients and validates well across multiple international data sources.

Unsupervised Learning for Automated Detection of Coronary Artery Disease Subgroups

Background The promise of precision population health includes the ability to use robust patient data to tailor prevention and care to specific groups. Advanced analytics may allow for automated detection of clinically informative subgroups that account for clinical, genetic, and environmental variability. This study sought to evaluate whether unsupervised machine learning approaches could interpret heterogeneous and missing clinical data to discover clinically important coronary artery disease subgroups. Methods and Results The Genetic Determinants of Peripheral Arterial Disease study is a prospective cohort that includes individuals with newly diagnosed and/or symptomatic coronary artery disease. We applied generalized low rank modeling and K-means cluster analysis using 155 phenotypic and genetic variables from 1329 participants. Cox proportional hazard models were used to examine associations between clusters and major adverse cardiovascular and cerebrovascular events and all-cause mortality. We then compared performance of risk stratification based on clusters and the American College of Cardiology/American Heart Association pooled cohort equations. Unsupervised analysis identified 4 phenotypically and prognostically distinct clusters. All-cause mortality was highest in cluster 1 (oldest/most comorbid; 26%), whereas major adverse cardiovascular and cerebrovascular event rates were highest in cluster 2 (youngest/multiethnic; 41%). Cluster 4 (middle-aged/healthiest behaviors) experienced more incident major adverse cardiovascular and cerebrovascular events (30%) than cluster 3 (middle-aged/lowest medication adherence; 23%), despite apparently similar risk factor and lifestyle profiles. In comparison with the pooled cohort equations, cluster membership was more informative for risk assessment of myocardial infarction, stroke, and mortality. Conclusions Unsupervised clustering identified 4 unique coronary artery disease subgroups with distinct clinical trajectories. Flexible unsupervised machine learning algorithms offer the ability to meaningfully process heterogeneous patient data and provide sharper insights into disease characterization and risk assessment. Registration URL: https://www.clinicaltrials.gov; Unique identifier: NCT00380185.

Computational drug repositioning of atorvastatin for ulcerative colitis

OBJECTIVE

Ulcerative colitis (UC) is a chronic inflammatory disorder with limited effective therapeutic options for long-term treatment and disease maintenance. We hypothesized that a multi-cohort analysis of independent cohorts representing real-world heterogeneity of UC would identify a robust transcriptomic signature to improve identification of FDA-approved drugs that can be repurposed to treat patients with UC.

MATERIALS AND METHODS

We performed a multi-cohort analysis of 272 colon biopsy transcriptome samples across 11 publicly available datasets to identify a robust UC disease gene signature. We compared the gene signature to in vitro transcriptomic profiles induced by 781 FDA-approved drugs to identify potential drug targets. We used a retrospective cohort study design modeled after a target trial to evaluate the protective effect of predicted drugs on colectomy risk in patients with UC from the Stanford Research Repository (STARR) database and Optum Clinformatics DataMart.

RESULTS

Atorvastatin treatment had the highest inverse-correlation with the UC gene signature among non-oncolytic FDA-approved therapies. In both STARR (n = 827) and Optum (n = 7821), atorvastatin intake was significantly associated with a decreased risk of colectomy, a marker of treatment-refractory disease, compared to patients prescribed a comparator drug (STARR: HR = 0.47, P = .03; Optum: HR = 0.66, P = .03), irrespective of age and length of atorvastatin treatment.

DISCUSSION & CONCLUSION

These findings suggest that atorvastatin may serve as a novel therapeutic option for ameliorating disease in patients with UC. Importantly, we provide a systematic framework for integrating publicly available heterogeneous molecular data with clinical data at a large scale to repurpose existing FDA-approved drugs for a wide range of human diseases.

An informatics consult approach for generating clinical evidence for treatment decisions

BACKGROUND

An Informatics Consult has been proposed in which clinicians request novel evidence from large scale health data resources, tailored to the treatment of a specific patient. However, the availability of such consultations is lacking. We seek to provide an Informatics Consult for a situation where a treatment indication and contraindication coexist in the same patient, i.e., anti-coagulation use for stroke prevention in a patient with both atrial fibrillation (AF) and liver cirrhosis.

METHODS

We examined four sources of evidence for the effect of warfarin on stroke risk or all-cause mortality from: (1) randomised controlled trials (RCTs), (2) meta-analysis of prior observational studies, (3) trial emulation (using population electronic health records (N = 3,854,710) and (4) genetic evidence (Mendelian randomisation). We developed prototype forms to request an Informatics Consult and return of results in electronic health record systems.

RESULTS

We found 0 RCT reports and 0 trials recruiting for patients with AF and cirrhosis. We found broad concordance across the three new sources of evidence we generated. Meta-analysis of prior observational studies showed that warfarin use was associated with lower stroke risk (hazard ratio [HR] = 0.71, CI 0.39-1.29). In a target trial emulation, warfarin was associated with lower all-cause mortality (HR = 0.61, CI 0.49-0.76) and ischaemic stroke (HR = 0.27, CI 0.08-0.91). Mendelian randomisation served as a drug target validation where we found that lower levels of vitamin K1 (warfarin is a vitamin K1 antagonist) are associated with lower stroke risk. A pilot survey with an independent sample of 34 clinicians revealed that 85% of clinicians found information on prognosis useful and that 79% thought that they should have access to the Informatics Consult as a service within their healthcare systems. We identified candidate steps for automation to scale evidence generation and to accelerate the return of results.

CONCLUSION

We performed a proof-of-concept Informatics Consult for evidence generation, which may inform treatment decisions in situations where there is dearth of randomised trials. Patients are surprised to know that their clinicians are currently not able to learn in clinic from data on 'patients like me'. We identify the key challenges in offering such an Informatics Consult as a service.

A survey of extant organizational and computational setups for deploying predictive models in health systems

OBJECTIVE

Artificial intelligence (AI) and machine learning (ML) enabled healthcare is now feasible for many health systems, yet little is known about effective strategies of system architecture and governance mechanisms for implementation. Our objective was to identify the different computational and organizational setups that early-adopter health systems have utilized to integrate AI/ML clinical decision support (AI-CDS) and scrutinize their trade-offs.

MATERIALS AND METHODS

We conducted structured interviews with health systems with AI deployment experience about their organizational and computational setups for deploying AI-CDS at point of care.

RESULTS

We contacted 34 health systems and interviewed 20 healthcare sites (58% response rate). Twelve (60%) sites used the native electronic health record vendor configuration for model development and deployment, making it the most common shared infrastructure. Nine (45%) sites used alternative computational configurations which varied significantly. Organizational configurations for managing AI-CDS were distinguished by how they identified model needs, built and implemented models, and were separable into 3 major types: Decentralized translation (n = 10, 50%), IT Department led (n = 2, 10%), and AI in Healthcare (AIHC) Team (n = 8, 40%).

DISCUSSION

No singular computational configuration enables all current use cases for AI-CDS. Health systems need to consider their desired applications for AI-CDS and whether investment in extending the off-the-shelf infrastructure is needed. Each organizational setup confers trade-offs for health systems planning strategies to implement AI-CDS.

CONCLUSION

Health systems will be able to use this framework to understand strengths and weaknesses of alternative organizational and computational setups when designing their strategy for artificial intelligence.

COVID-19 in patients with autoimmune diseases: characteristics and outcomes in a multinational network of cohorts across three countries

OBJECTIVE

Patients with autoimmune diseases were advised to shield to avoid coronavirus disease 2019 (COVID-19), but information on their prognosis is lacking. We characterized 30-day outcomes and mortality after hospitalization with COVID-19 among patients with prevalent autoimmune diseases, and compared outcomes after hospital admissions among similar patients with seasonal influenza.

METHODS

A multinational network cohort study was conducted using electronic health records data from Columbia University Irving Medical Center [USA, Optum (USA), Department of Veterans Affairs (USA), Information System for Research in Primary Care-Hospitalization Linked Data (Spain) and claims data from IQVIA Open Claims (USA) and Health Insurance and Review Assessment (South Korea). All patients with prevalent autoimmune diseases, diagnosed and/or hospitalized between January and June 2020 with COVID-19, and similar patients hospitalized with influenza in 2017-18 were included. Outcomes were death and complications within 30 days of hospitalization.

RESULTS

We studied 133 589 patients diagnosed and 48 418 hospitalized with COVID-19 with prevalent autoimmune diseases. Most patients were female, aged ≥50 years with previous comorbidities. The prevalence of hypertension (45.5-93.2%), chronic kidney disease (14.0-52.7%) and heart disease (29.0-83.8%) was higher in hospitalized vs diagnosed patients with COVID-19. Compared with 70 660 hospitalized with influenza, those admitted with COVID-19 had more respiratory complications including pneumonia and acute respiratory distress syndrome, and higher 30-day mortality (2.2-4.3% vs 6.32-24.6%).

CONCLUSION

Compared with influenza, COVID-19 is a more severe disease, leading to more complications and higher mortality.

Characteristics and Outcomes of Over 300,000 Patients with COVID-19 and History of Cancer in the United States and Spain

BACKGROUND

We described the demographics, cancer subtypes, comorbidities, and outcomes of patients with a history of cancer and coronavirus disease 2019 (COVID-19). Second, we compared patients hospitalized with COVID-19 to patients diagnosed with COVID-19 and patients hospitalized with influenza.

METHODS

We conducted a cohort study using eight routinely collected health care databases from Spain and the United States, standardized to the Observational Medical Outcome Partnership common data model. Three cohorts of patients with a history of cancer were included: (i) diagnosed with COVID-19, (ii) hospitalized with COVID-19, and (iii) hospitalized with influenza in 2017 to 2018. Patients were followed from index date to 30 days or death. We reported demographics, cancer subtypes, comorbidities, and 30-day outcomes.

RESULTS

We included 366,050 and 119,597 patients diagnosed and hospitalized with COVID-19, respectively. Prostate and breast cancers were the most frequent cancers (range: 5%-18% and 1%-14% in the diagnosed cohort, respectively). Hematologic malignancies were also frequent, with non-Hodgkin's lymphoma being among the five most common cancer subtypes in the diagnosed cohort. Overall, patients were aged above 65 years and had multiple comorbidities. Occurrence of death ranged from 2% to 14% and from 6% to 26% in the diagnosed and hospitalized COVID-19 cohorts, respectively. Patients hospitalized with influenza (n = 67,743) had a similar distribution of cancer subtypes, sex, age, and comorbidities but lower occurrence of adverse events.

CONCLUSIONS

Patients with a history of cancer and COVID-19 had multiple comorbidities and a high occurrence of COVID-19-related events. Hematologic malignancies were frequent.

IMPACT

This study provides epidemiologic characteristics that can inform clinical care and etiologic studies.

Thirty-Day Outcomes of Children and Adolescents With COVID-19: An International Experience

OBJECTIVES

To characterize the demographics, comorbidities, symptoms, in-hospital treatments, and health outcomes among children and adolescents diagnosed or hospitalized with coronavirus disease 2019 (COVID-19) and to compare them in secondary analyses with patients diagnosed with previous seasonal influenza in 2017-2018.

METHODS

International network cohort using real-world data from European primary care records (France, Germany, and Spain), South Korean claims and US claims, and hospital databases. We included children and adolescents diagnosed and/or hospitalized with COVID-19 at age <18 between January and June 2020. We described baseline demographics, comorbidities, symptoms, 30-day in-hospital treatments, and outcomes including hospitalization, pneumonia, acute respiratory distress syndrome, multisystem inflammatory syndrome in children, and death.

RESULTS

A total of 242 158 children and adolescents diagnosed and 9769 hospitalized with COVID-19 and 2 084 180 diagnosed with influenza were studied. Comorbidities including neurodevelopmental disorders, heart disease, and cancer were more common among those hospitalized with versus diagnosed with COVID-19. Dyspnea, bronchiolitis, anosmia, and gastrointestinal symptoms were more common in COVID-19 than influenza. In-hospital prevalent treatments for COVID-19 included repurposed medications (<10%) and adjunctive therapies: systemic corticosteroids (6.8%-7.6%), famotidine (9.0%-28.1%), and antithrombotics such as aspirin (2.0%-21.4%), heparin (2.2%-18.1%), and enoxaparin (2.8%-14.8%). Hospitalization was observed in 0.3% to 1.3% of the cohort diagnosed with COVID-19, with undetectable (n < 5 per database) 30-day fatality. Thirty-day outcomes including pneumonia and hypoxemia were more frequent in COVID-19 than influenza.

CONCLUSIONS

Despite negligible fatality, complications including hospitalization, hypoxemia, and pneumonia were more frequent in children and adolescents with COVID-19 than with influenza. Dyspnea, anosmia, and gastrointestinal symptoms could help differentiate diagnoses. A wide range of medications was used for the inpatient management of pediatric COVID-19.

Treatment and Monitoring Variability in US Metastatic Breast Cancer Care

PURPOSE

Treatment and monitoring options for patients with metastatic breast cancer (MBC) are increasing, but little is known about variability in care. We sought to improve understanding of MBC care and its correlates by analyzing real-world claims data using a search engine with a novel query language to enable temporal electronic phenotyping.

METHODS

Using the Advanced Cohort Engine, we identified 6,180 women who met criteria for having estrogen receptor-positive, human epidermal growth factor receptor 2-negative MBC from IBM MarketScan US insurance claims (2007-2014). We characterized treatment, monitoring, and hospice usage, along with clinical and nonclinical factors affecting care.

RESULTS

We observed wide variability in treatment modality and monitoring across patients and geography. Most women received first-recorded therapy with endocrine (67%) versus chemotherapy, underwent more computed tomography (CT) (76%) than positron emission tomography-CT, and were monitored using tumor markers (58%). Nearly half (46%) met criteria for aggressive disease, which were associated with receiving chemotherapy first, monitoring primarily with CT, and more frequent imaging. Older age was associated with endocrine therapy first, less frequent imaging, and less use of tumor markers. After controlling for clinical factors, care strategies varied significantly by nonclinical factors (median regional income with first-recorded therapy and imaging type, geographic region with these and with imaging frequency and use of tumor markers; P < .0001).

CONCLUSION

Variability in US MBC care is explained by patient and disease factors and by nonclinical factors such as geographic region, suggesting that treatment decisions are influenced by local practice patterns and/or resources. A search engine designed to express complex electronic phenotypes from longitudinal patient records enables the identification of variability in patient care, helping to define disparities and areas for improvement.

Automated model versus treating physician for predicting survival time of patients with metastatic cancer

OBJECTIVE

Being able to predict a patient's life expectancy can help doctors and patients prioritize treatments and supportive care. For predicting life expectancy, physicians have been shown to outperform traditional models that use only a few predictor variables. It is possible that a machine learning model that uses many predictor variables and diverse data sources from the electronic medical record can improve on physicians' performance. For patients with metastatic cancer, we compared accuracy of life expectancy predictions by the treating physician, a machine learning model, and a traditional model.

MATERIALS AND METHODS

A machine learning model was trained using 14 600 metastatic cancer patients' data to predict each patient's distribution of survival time. Data sources included note text, laboratory values, and vital signs. From 2015-2016, 899 patients receiving radiotherapy for metastatic cancer were enrolled in a study in which their radiation oncologist estimated life expectancy. Survival predictions were also made by the machine learning model and a traditional model using only performance status. Performance was assessed with area under the curve for 1-year survival and calibration plots.

RESULTS

The radiotherapy study included 1190 treatment courses in 899 patients. A total of 879 treatment courses in 685 patients were included in this analysis. Median overall survival was 11.7 months. Physicians, machine learning model, and traditional model had area under the curve for 1-year survival of 0.72 (95% CI 0.63-0.81), 0.77 (0.73-0.81), and 0.68 (0.65-0.71), respectively.

CONCLUSIONS

The machine learning model's predictions were more accurate than those of the treating physician or a traditional model.

Learning decision thresholds for risk stratification models from aggregate clinician behavior

Using a risk stratification model to guide clinical practice often requires the choice of a cutoff—called the decision threshold—on the model’s output to trigger a subsequent action such as an electronic alert. Choosing this cutoff is not always straightforward. We propose a flexible approach that leverages the collective information in treatment decisions made in real life to learn reference decision thresholds from physician practice. Using the example of prescribing a statin for primary prevention of cardiovascular disease based on 10-year risk calculated by the 2013 pooled cohort equations, we demonstrate the feasibility of using real-world data to learn the implicit decision threshold that reflects existing physician behavior. Learning a decision threshold in this manner allows for evaluation of a proposed operating point against the threshold reflective of the community standard of care. Furthermore, this approach can be used to monitor and audit model-guided clinical decision making following model deployment.

Characteristics and outcomes of 627 044 COVID-19 patients living with and without obesity in the United States, Spain, and the United Kingdom

BACKGROUND

A detailed characterization of patients with COVID-19 living with obesity has not yet been undertaken. We aimed to describe and compare the demographics, medical conditions, and outcomes of COVID-19 patients living with obesity (PLWO) to those of patients living without obesity.

METHODS

We conducted a cohort study based on outpatient/inpatient care and claims data from January to June 2020 from Spain, the UK, and the US. We used six databases standardized to the OMOP common data model. We defined two non-mutually exclusive cohorts of patients diagnosed and/or hospitalized with COVID-19; patients were followed from index date to 30 days or death. We report the frequency of demographics, prior medical conditions, and 30-days outcomes (hospitalization, events, and death) by obesity status.

RESULTS

We included 627 044 (Spain: 122 058, UK: 2336, and US: 502 650) diagnosed and 160 013 (Spain: 18 197, US: 141 816) hospitalized patients with COVID-19. The prevalence of obesity was higher among patients hospitalized (39.9%, 95%CI: 39.8-40.0) than among those diagnosed with COVID-19 (33.1%; 95%CI: 33.0-33.2). In both cohorts, PLWO were more often female. Hospitalized PLWO were younger than patients without obesity. Overall, COVID-19 PLWO were more likely to have prior medical conditions, present with cardiovascular and respiratory events during hospitalization, or require intensive services compared to COVID-19 patients without obesity.

CONCLUSION

We show that PLWO differ from patients without obesity in a wide range of medical conditions and present with more severe forms of COVID-19, with higher hospitalization rates and intensive services requirements. These findings can help guiding preventive strategies of COVID-19 infection and complications and generating hypotheses for causal inference studies.

ACE: the Advanced Cohort Engine for searching longitudinal patient records

OBJECTIVE

To propose a paradigm for a scalable time-aware clinical data search, and to describe the design, implementation and use of a search engine realizing this paradigm.

MATERIALS AND METHODS

The Advanced Cohort Engine (ACE) uses a temporal query language and in-memory datastore of patient objects to provide a fast, scalable, and expressive time-aware search. ACE accepts data in the Observational Medicine Outcomes Partnership Common Data Model, and is configurable to balance performance with compute cost. ACE's temporal query language supports automatic query expansion using clinical knowledge graphs. The ACE API can be used with R, Python, Java, HTTP, and a Web UI.

RESULTS

ACE offers an expressive query language for complex temporal search across many clinical data types with multiple output options. ACE enables electronic phenotyping and cohort-building with subsecond response times in searching the data of millions of patients for a variety of use cases.

DISCUSSION

ACE enables fast, time-aware search using a patient object-centric datastore, thereby overcoming many technical and design shortcomings of relational algebra-based querying. Integrating electronic phenotype development with cohort-building enables a variety of high-value uses for a learning health system. Tradeoffs include the need to learn a new query language and the technical setup burden.

CONCLUSION

ACE is a tool that combines a unique query language for time-aware search of longitudinal patient records with a patient object datastore for rapid electronic phenotyping, cohort extraction, and exploratory data analyses.

Improving hospital readmission prediction using individualized utility analysis

OBJECTIVE

Machine learning (ML) models for allocating readmission-mitigating interventions are typically selected according to their discriminative ability, which may not necessarily translate into utility in allocation of resources. Our objective was to determine whether ML models for allocating readmission-mitigating interventions have different usefulness based on their overall utility and discriminative ability.

MATERIALS AND METHODS

We conducted a retrospective utility analysis of ML models using claims data acquired from the Optum Clinformatics Data Mart, including 513,495 commercially-insured inpatients (mean [SD] age 69 [19] years; 294,895 [57%] Female) over the period January 2016 through January 2017 from all 50 states with mean 90 day cost of $11,552. Utility analysis estimates the cost, in dollars, of allocating interventions for lowering readmission risk based on the reduction in the 90-day cost.

RESULTS

Allocating readmission-mitigating interventions based on a GBDT model trained to predict readmissions achieved an estimated utility gain of $104 per patient, and an AUC of 0.76 (95% CI 0.76, 0.77); allocating interventions based on a model trained to predict cost as a proxy achieved a higher utility of $175.94 per patient, and an AUC of 0.62 (95% CI 0.61, 0.62). A hybrid model combining both intervention strategies is comparable with the best models on either metric. Estimated utility varies by intervention cost and efficacy, with each model performing the best under different intervention settings.

CONCLUSION

We demonstrate that machine learning models may be ranked differently based on overall utility and discriminative ability. Machine learning models for allocation of limited health resources should consider directly optimizing for utility.

Use of repurposed and adjuvant drugs in hospital patients with covid-19: multinational network cohort study

OBJECTIVE

To investigate the use of repurposed and adjuvant drugs in patients admitted to hospital with covid-19 across three continents.

DESIGN

Multinational network cohort study.

SETTING

Hospital electronic health records from the United States, Spain, and China, and nationwide claims data from South Korea.

PARTICIPANTS

303 264 patients admitted to hospital with covid-19 from January 2020 to December 2020.

MAIN OUTCOME MEASURES

Prescriptions or dispensations of any drug on or 30 days after the date of hospital admission for covid-19.

RESULTS

Of the 303 264 patients included, 290 131 were from the US, 7599 from South Korea, 5230 from Spain, and 304 from China. 3455 drugs were identified. Common repurposed drugs were hydroxychloroquine (used in from <5 (<2%) patients in China to 2165 (85.1%) in Spain), azithromycin (from 15 (4.9%) in China to 1473 (57.9%) in Spain), combined lopinavir and ritonavir (from 156 (<2%) in the VA-OMOP US to 2,652 (34.9%) in South Korea and 1285 (50.5%) in Spain), and umifenovir (0% in the US, South Korea, and Spain and 238 (78.3%) in China). Use of adjunctive drugs varied greatly, with the five most used treatments being enoxaparin, fluoroquinolones, ceftriaxone, vitamin D, and corticosteroids. Hydroxychloroquine use increased rapidly from March to April 2020 but declined steeply in May to June and remained low for the rest of the year. The use of dexamethasone and corticosteroids increased steadily during 2020.

CONCLUSIONS

Multiple drugs were used in the first few months of the covid-19 pandemic, with substantial geographical and temporal variation. Hydroxychloroquine, azithromycin, lopinavir-ritonavir, and umifenovir (in China only) were the most prescribed repurposed drugs. Antithrombotics, antibiotics, H2 receptor antagonists, and corticosteroids were often used as adjunctive treatments. Research is needed on the comparative risk and benefit of these treatments in the management of covid-19.

MINIMAR (MINimum Information for Medical AI Reporting): Developing reporting standards for artificial intelligence in health care

The rise of digital data and computing power have contributed to significant advancements in artificial intelligence (AI), leading to the use of classification and prediction models in health care to enhance clinical decision-making for diagnosis, treatment and prognosis. However, such advances are limited by the lack of reporting standards for the data used to develop those models, the model architecture, and the model evaluation and validation processes. Here, we present MINIMAR (MINimum Information for Medical AI Reporting), a proposal describing the minimum information necessary to understand intended predictions, target populations, and hidden biases, and the ability to generalize these emerging technologies. We call for a standard to accurately and responsibly report on AI in health care. This will facilitate the design and implementation of these models and promote the development and use of associated clinical decision support tools, as well as manage concerns regarding accuracy and bias.

Ontology-driven weak supervision for clinical entity classification in electronic health records

In the electronic health record, using clinical notes to identify entities such as disorders and their temporality (e.g. the order of an event relative to a time index) can inform many important analyses. However, creating training data for clinical entity tasks is time consuming and sharing labeled data is challenging due to privacy concerns. The information needs of the COVID-19 pandemic highlight the need for agile methods of training machine learning models for clinical notes. We present Trove, a framework for weakly supervised entity classification using medical ontologies and expert-generated rules. Our approach, unlike hand-labeled notes, is easy to share and modify, while offering performance comparable to learning from manually labeled training data. In this work, we validate our framework on six benchmark tasks and demonstrate Trove's ability to analyze the records of patients visiting the emergency department at Stanford Health Care for COVID-19 presenting symptoms and risk factors.

Development and validation of phenotype classifiers across multiple sites in the observational health data sciences and informatics network

Objective

Accurate electronic phenotyping is essential to support collaborative observational research. Supervised machine learning methods can be used to train phenotype classifiers in a high-throughput manner using imperfectly labeled data. We developed 10 phenotype classifiers using this approach and evaluated performance across multiple sites within the Observational Health Data Sciences and Informatics (OHDSI) network.

Materials and Methods

We constructed classifiers using the Automated PHenotype Routine for Observational Definition, Identification, Training and Evaluation (APHRODITE) R-package, an open-source framework for learning phenotype classifiers using datasets in the Observational Medical Outcomes Partnership Common Data Model. We labeled training data based on the presence of multiple mentions of disease-specific codes. Performance was evaluated on cohorts derived using rule-based definitions and real-world disease prevalence. Classifiers were developed and evaluated across 3 medical centers, including 1 international site.

Results

Compared to the multiple mentions labeling heuristic, classifiers showed a mean recall boost of 0.43 with a mean precision loss of 0.17. Performance decreased slightly when classifiers were shared across medical centers, with mean recall and precision decreasing by 0.08 and 0.01, respectively, at a site within the USA, and by 0.18 and 0.10, respectively, at an international site.

Discussion and Conclusion

We demonstrate a high-throughput pipeline for constructing and sharing phenotype classifiers across sites within the OHDSI network using APHRODITE. Classifiers exhibit good portability between sites within the USA, however limited portability internationally, indicating that classifier generalizability may have geographic limitations, and, consequently, sharing the classifier-building recipe, rather than the pretrained classifiers, may be more useful for facilitating collaborative observational research.

Assessment of Extractability and Accuracy of Electronic Health Record Data for Joint Implant Registries

IMPORTANCE

Implant registries provide valuable information on the performance of implants in a real-world setting, yet they have traditionally been expensive to establish and maintain. Electronic health records (EHRs) are widely used and may include the information needed to generate clinically meaningful reports similar to a formal implant registry.

OBJECTIVES

To quantify the extractability and accuracy of registry-relevant data from the EHR and to assess the ability of these data to track trends in implant use and the durability of implants (hereafter referred to as implant survivorship), using data stored since 2000 in the EHR of the largest integrated health care system in the United States.

DESIGN, SETTING, AND PARTICIPANTS

Retrospective cohort study of a large EHR of veterans who had 45 351 total hip arthroplasty procedures in Veterans Health Administration hospitals from 2000 to 2017. Data analysis was performed from January 1, 2000, to December 31, 2017.

EXPOSURES

Total hip arthroplasty.

MAIN OUTCOMES AND MEASURES

Number of total hip arthroplasty procedures extracted from the EHR, trends in implant use, and relative survivorship of implants.

RESULTS

A total of 45 351 total hip arthroplasty procedures were identified from 2000 to 2017 with 192 805 implant parts. Data completeness improved over the time. After 2014, 85% of prosthetic heads, 91% of shells, 81% of stems, and 85% of liners used in the Veterans Health Administration health care system were identified by part number. Revision burden and trends in metal vs ceramic prosthetic femoral head use were found to reflect data from the American Joint Replacement Registry. Recalled implants were obvious negative outliers in implant survivorship using Kaplan-Meier curves.

CONCLUSIONS AND RELEVANCE

Although loss to follow-up remains a challenge that requires additional attention to improve the quantitative nature of calculated implant survivorship, we conclude that data collected during routine clinical care and stored in the EHR of a large health system over 18 years were sufficient to provide clinically meaningful data on trends in implant use and to identify poor implants that were subsequently recalled. This automated approach was low cost and had no reporting burden. This low-cost, low-overhead method to assess implant use and performance within a large health care setting may be useful to internal quality assurance programs and, on a larger scale, to postmarket surveillance of implant performance.

The number needed to benefit: estimating the value of predictive analytics in healthcare

Predictive analytics in health care has generated increasing enthusiasm recently, as reflected in a rapidly growing body of predictive models reported in literature and in real-time embedded models using electronic health record data. However, estimating the benefit of applying any single model to a specific clinical problem remains challenging today. Developing a shared framework for estimating model value is therefore critical to facilitate the effective, safe, and sustainable use of predictive tools into the future. We highlight key concepts within the prediction-action dyad that together are expected to impact model benefit. These include factors relevant to model prediction (including the number needed to screen) as well as those relevant to the subsequent action (number needed to treat). In the simplest terms, a number needed to benefit contextualizes the numbers needed to screen and treat, offering an opportunity to estimate the value of a clinical predictive model in action.