Mahesh Vaidyanathan

Mahesh Vaidyanathan

Auto-generated profile, based on publicly available information

Mahesh Vaidyanathan

Northwestern Memorial HealthCare

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
31 May 2022 in JMIR Medical Informatics
Verified
Journal Article
Open access
JHJonathan Huang
Jonathan Huang
GGGalal Galal
Galal Galal
MEMozziyar Etemadi
Mozziyar Etemadi
MVMahesh Vaidyanathan
Mahesh Vaidyanathan
Background: Racial bias is a key concern regarding the development, validation, and implementation of machine learning (ML) models in clinical settings. Despite the potential of bias to propagate health disparities, racial bias in clinical ML has yet to be thoroughly examined and best practices for bias mitigation remain unclear. Objective: Our objective was to perform a scoping review to characterize the methods by which the racial bias of ML has been assessed and describe strategies that may be used to enhance algorithmic fairness in clinical ML. Methods: A scoping review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Extension for Scoping Reviews. A literature search using PubMed, Scopus, and Embase databases, as well as Google Scholar, identified 635 records, of which 12 studies were included. Results: Applications of ML were varied and involved diagnosis, outcome prediction, and clinical score prediction performed on data sets including images, diagnostic studies, clinical text, and clinical variables. Of the 12 studies, 1 (8%) described a model in routine clinical use, 2 (17%) examined prospectively validated clinical models, and the remaining 9 (75%) described internally validated models. In addition, 8 (67%) studies concluded that racial bias was present, 2 (17%) concluded that it was not, and 2 (17%) assessed the implementation of bias mitigation strategies without comparison to a baseline model. Fairness metrics used to assess algorithmic racial bias were inconsistent. The most commonly observed metrics were equal opportunity difference (5/12, 42%), accuracy (4/12, 25%), and disparate impact (2/12, 17%). All 8 (67%) studies that implemented methods for mitigation of racial bias successfully increased fairness, as measured by the authors’ chosen metrics. Preprocessing methods of bias mitigation were most commonly used across all studies that implemented them. Conclusions: The broad scope of medical ML applications and potential patient harms demand an increased emphasis on evaluation and mitigation of racial bias in clinical ML. However, the adoption of algorithmic fairness principles in medicine remains inconsistent and is limited by poor data availability and ML model reporting. We recommend that researchers and journal editors emphasize standardized reporting and data availability in medical ML studies to improve transparency and facilitate evaluation for racial bias.

29 January 2025 in JMIR Formative Research
Verified
Journal Article
Open access
JKJodie Koh
Jodie Koh
SCStacey Caron
Stacey Caron
AWAmber N Watters
Amber N Watters
MVMahesh Vaidyanathan
Mahesh Vaidyanathan
DMDavid Melnick
David Melnick
MEMozziyar Etemadi
Mozziyar Etemadi
Background: Patient recruitment and data management are laborious, resource-intensive aspects of clinical research that often dictate whether the successful completion of studies is possible. Technological advances present opportunities for streamlining these processes, thus improving completion rates for clinical research studies. Objective: This paper aims to demonstrate how technological adjuncts can enhance clinical research processes via automation and digital integration. Methods: Using one clinical research study as an example, we highlighted the use of technological adjuncts to automate and streamline research processes across various digital platforms, including a centralized database of electronic medical records (enterprise data warehouse [EDW]); a clinical research data management tool (REDCap [Research Electronic Data Capture]); and a locally managed, Health Insurance Portability and Accountability Act–compliant server. Eligible participants were identified through automated queries in the EDW, after which they received personalized email invitations with digital consent forms. After digital consent, patient data were transferred to a single Health Insurance Portability and Accountability Act–compliant server where each participant was assigned a unique QR code to facilitate data collection and integration. After the research study visit, data obtained were associated with existing electronic medical record data for each participant via a QR code system that collated participant consent, imaging data, and associated clinical data according to a unique examination ID. Results: Over a 19-month period, automated EDW queries identified 20,988 eligible patients, and 10,582 patients received personalized email invitations. In total, 1000 (9.45%) patients signed consents to participate in the study. Of the consented patients, 549 unique patients completed 779 study visits; some patients consented to the study at more than 1 time period during their pregnancy. Conclusions: Technological adjuncts in clinical research decrease human labor while increasing participant reach and minimizing disruptions to clinic operations. Automating portions of the clinical research process benefits clinical research efforts by expanding and optimizing participant reach while reducing the limitations of labor and time in completing research studies.

12 January 2022 in JMIR Preprints
Verified
Preprint
Open access
JHJonathan Huang
Jonathan Huang
GGGalal Galal
Galal Galal
MEMozziyar Etemadi
Mozziyar Etemadi
MVMahesh Vaidyanathan
Mahesh Vaidyanathan
BACKGROUND Racial bias is a key concern regarding the development, validation, and implementation of machine learning (ML) models in clinical settings. Despite the potential of bias to propagate health disparities, racial bias in clinical ML has yet to be thoroughly examined and best practices for bias mitigation remain unclear. OBJECTIVE Our objective was to perform a scoping review to characterize the methods by which the racial bias of ML has been assessed and describe strategies that may be used to enhance algorithmic fairness in clinical ML. METHODS A scoping review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Extension for Scoping Reviews. A literature search using PubMed, Scopus, and Embase databases, as well as Google Scholar, identified 635 records, of which 12 studies were included. RESULTS Applications of ML were varied and involved diagnosis, outcome prediction, and clinical score prediction performed on data sets including images, diagnostic studies, clinical text, and clinical variables. Of the 12 studies, 1 (8%) described a model in routine clinical use, 2 (17%) examined prospectively validated clinical models, and the remaining 9 (75%) described internally validated models. In addition, 8 (67%) studies concluded that racial bias was present, 2 (17%) concluded that it was not, and 2 (17%) assessed the implementation of bias mitigation strategies without comparison to a baseline model. Fairness metrics used to assess algorithmic racial bias were inconsistent. The most commonly observed metrics were equal opportunity difference (5/12, 42%), accuracy (4/12, 25%), and disparate impact (2/12, 17%). All 8 (67%) studies that implemented methods for mitigation of racial bias successfully increased fairness, as measured by the authors’ chosen metrics. Preprocessing methods of bias mitigation were most commonly used across all studies that implemented them. CONCLUSIONS The broad scope of medical ML applications and potential patient harms demand an increased emphasis on evaluation and mitigation of racial bias in clinical ML. However, the adoption of algorithmic fairness principles in medicine remains inconsistent and is limited by poor data availability and ML model reporting. We recommend that researchers and journal editors emphasize standardized reporting and data availability in medical ML studies to improve transparency and facilitate evaluation for racial bias.

12 April 2024 in JMIR Preprints
Verified
Preprint
Open access
JKJodie Koh
Jodie Koh
SCStacey Caron
Stacey Caron
AWAmber N Watters
Amber N Watters
MVMahesh Vaidyanathan
Mahesh Vaidyanathan
DMDavid Melnick
David Melnick
MEMozziyar Etemadi
Mozziyar Etemadi
BACKGROUND Patient recruitment and data management are laborious, resource-intensive aspects of clinical research that often dictate whether the successful completion of studies is possible. Technological advances present opportunities for streamlining these processes, thus improving completion rates for clinical research studies. OBJECTIVE This paper aims to demonstrate how technological adjuncts can enhance clinical research processes via automation and digital integration. METHODS Using one clinical research study as an example, we highlighted the use of technological adjuncts to automate and streamline research processes across various digital platforms, including a centralized database of electronic medical records (enterprise data warehouse [EDW]); a clinical research data management tool (REDCap [Research Electronic Data Capture]); and a locally managed, Health Insurance Portability and Accountability Act–compliant server. Eligible participants were identified through automated queries in the EDW, after which they received personalized email invitations with digital consent forms. After digital consent, patient data were transferred to a single Health Insurance Portability and Accountability Act–compliant server where each participant was assigned a unique QR code to facilitate data collection and integration. After the research study visit, data obtained were associated with existing electronic medical record data for each participant via a QR code system that collated participant consent, imaging data, and associated clinical data according to a unique examination ID. RESULTS Over a 19-month period, automated EDW queries identified 20,988 eligible patients, and 10,582 patients received personalized email invitations. In total, 1000 (9.45%) patients signed consents to participate in the study. Of the consented patients, 549 unique patients completed 779 study visits; some patients consented to the study at more than 1 time period during their pregnancy. CONCLUSIONS Technological adjuncts in clinical research decrease human labor while increasing participant reach and minimizing disruptions to clinic operations. Automating portions of the clinical research process benefits clinical research efforts by expanding and optimizing participant reach while reducing the limitations of labor and time in completing research studies.

27 October 2026 in Advances in Cancer Education and Quality Improvement
Verified
Journal Article
Open access
PDF file
SDShahin Davoudpour
Shahin Davoudpour
MVMahesh Vaidyanathan
Mahesh Vaidyanathan
RBRinad Beidas
Rinad Beidas
HMHannah Mulroe
Hannah Mulroe
EBElliot Brandon
Elliot Brandon
GAGaurava Agarwal
Gaurava Agarwal
Objective: To evaluate an integrated machine learning model using routinely collected organizational data to identify nurses at increased risk for burnout. Background: Health care worker burnout directly threatens patient safety, quality of care, and workforce stability. Although nurses experience significantly higher burnout rates than physicians and face distinct operational stressors (eg, staffing shortages, continuous bedside care demands), most predictive modeling efforts have focused on physicians. Current approaches to identifying burnout typically rely on voluntary surveys, but advances in data science and machine learning have created new opportunities to move beyond self-reported measures. Methods: This retrospective observational study included 1,177 nurses from a large health system in the Midwest. We integrated multiple routinely collected administrative sources to create a comprehensive dataset that included information from the electronic health record, human resources, and annual health system surveys. We developed a machine learning model using the XGBoost (extreme gradient boosting) framework. The model was trained using 80% of the sample and evaluated in the remaining 20% of the test sample (n =236). Model performance was assessed using standard classification metrics. The target variable was self-reported burnout assessed using a validated 2-item instrument measuring emotional exhaustion and depersonalization. Results: The machine learning model demonstrated strong performance in identifying nurses who met the criteria for burnout. The model achieved an area-under-the-receiver-operating-characteristic curve of 0.81, indicating good discrimination between nurses with and without burnout. The model identified 81.3% (bootstrap 95% CI = 71.3%–88.3%) of nurses were experiencing burnout (sensitivity) while maintaining a positive predictive value of 65.0% (bootstrap 95% CI = 55.3%–73.6%), and F1 score of 0.72 (bootstrap 95% CI = 0.64–0.79). Conclusion: Using integrated data sources, we developed a model that demonstrated strong performance in prospectively identifying nurses who were experiencing burnout. This is the first step to providing health systems with an actionable foundation to implement targeted, system-level workforce interventions.