KNKeshav Nandakumar

Keshav Nandakumar

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Keshav Nandakumar has a total of 26 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Nighttime Interruptions and Clinical Risk: A Systematized Review of After-hours Paging and Error-related Outcomes

    by Connor D. Yost, Hannah D’CUnha, Noel Balli, Ron Tang, Keshav Nandakumar, Claudia J. Garcia, Evan Jacobs, Paul Tran, Laura Mercer, Tamlyn Tarzia, et al.

    Nighttime inpatient cross-cover combines fatigue, reduced staffing, limited patient familiarity, and frequent paging interruptions, with resident physicians carrying much of this workload in teaching hospitals. We conducted a systematized PubMed review (search executed January 2026; publications from January 2021 to January 2026) using a 4-concept search covering night or after-hours work, inpatient settings, interruptions (pages, alerts, secure messages), and medical errors or patient-safety outcomes. Each of 112 records was independently screened by all 5 reviewers against a priori eligibility criteria, with disagreements resolved by consensus. Of these, 45 were excluded at title/abstract and 48 at full text; 19 studies met the inclusion criteria. Designs were quality improvement [n=8 (42%)], interview or survey [n=6 (32%)], and observational or retrospective [n=5 (26%)]; most were academic teaching hospitals in North America. Overnight interruptions most often involved medication questions, plan-of-care clarifications, handoff-related communication, diagnostic workflow disruptions, or electronic alerts. Most outcomes were process measures or error surrogates rather than direct clinical endpoints. Interventions that standardized communication or restructured notification workflows consistently reduced interruption burden, improved handoff efficiency, or increased capture of clinically significant overnight events. Combined with fatigue, reduced staffing, and limited continuity, nighttime interruptions amplify cognitive load and increase vulnerability to delays, omissions, and unsafe cross-cover. The most consistent benefits came from structured handoffs, clearer escalation expectations, and strategies that reduce deferrable or ambiguous overnight communication. Multicenter studies linking paging events to downstream EHR actions and patient-centered outcomes are needed.

  • Open Access

    Retrieval-Augmented Large Language Model for Angiographic Prediction of Coronary Physiology

    by Sant Kumar, Keshav Nandakumar, Pedro A. Villablanca, Vikas Aggarwal, Hursh Naik, Nezar Falluji

    Background: Invasive physiologic assessment is often needed for moderately stenotic coronary lesions, although it remains underused in routine practice. It remains unknown whether GPT-based large language models (LLMs) can estimate coronary physiology from coronary angiographic images, and whether retrieval augmentation can improve this task. Methods: We performed a retrospective pilot study of consecutive cases undergoing coronary angiography with invasive instantaneous wave-free ratio (iFR) assessment between 2023 and 2025. Eligible cases required invasive iFR and two orthogonal end-diastolic still frames of the target vessel at maximal opacification. We compared a baseline GPT-5.2 model without retrieval-augmented generation (RAG), termed No-RAG, with the same GPT-5.2 model using RAG, termed RAG. Both conditions received identical angiographic frames and structured clinical text. The RAG modification added the top five case-specific text chunks retrieved from four coronary physiology/revascularization documents to provide physiologic thresholds, guideline context, and uncertainty framing; no additional angiographic images or lesion-specific iFR information were provided. Frame-level predictions were averaged to derive a case-level predicted iFR. The primary endpoint was agreement between predicted and measured iFR. Results: Of 34 eligible cases screened, 32 vessels were included. The cohort comprised 25/32 cases (78.1%) with significant disease (iFR ≤ 0.89) and 7 cases classified as non-ischemic. Without RAG, mean absolute error (MAE) was 0.064 and the root mean square error (RMSE) was 0.083, with weak correlation with invasive iFR (r = 0.205, p = 0.259). With RAG, point estimates favored improved continuous agreement, with MAE decreasing to 0.029, RMSE to 0.038, and correlation increasing to r = 0.830 (p < 0.001). Threshold-based classification also yielded higher point estimates for accuracy, increasing from 0.750 to 0.906. Conclusions: In this small pilot study, improved point estimates for agreement between LLM-predicted and invasively measured iFR were seen after adding RAG to a GPT-based model for estimating iFR from angiographic imaging. These findings suggest that functionally classifying coronary stenoses is limited by overestimating severity in less severe stenoses, and that a scaling correction is needed. The results, however, require validation in larger, more balanced cohorts.