Extreme Clinical Code Prediction
Graph-constrained learning for accurate, hierarchy-aware ICD-10-CM code prediction from clinical notes.
Explore projectBukhari Lab · Biomedical Data Science & Healthcare AI
Advancing Biomedical Data Science through Transparent, Explainable, and Trustworthy AI
HUMAN + AI > HUMAN
Skyline: King of Hearts / Wikimedia Commons, CC BY-SA 3.0. Cropped and resized.
Current research
Active Projects
Recent initiatives spanning trustworthy clinical AI, interoperability, predictive analytics, and biomedical discovery.
FHIRTrustBench
A benchmark for evaluating the readiness, interoperability, and trustworthiness of FHIR-based clinical AI systems.
Explore projectCare Escalation Prediction in the ICU
Early multi-label prediction of respiratory, hemodynamic, renal, and neurological deterioration from EHR data.
Explore projectPathway-Driven Therapeutic Target Discovery
A machine-learning framework that uses biological pathways to identify therapeutic targets in metabolic disorders.
Explore projectClinician Burnout Surveillance
Narrative-driven analysis of ICU clinical notes to surface actionable signals related to clinician burnout risk.
Explore projectRecent scholarship
Featured Papers
Selected recent publications and preprints from the lab's current research program.
arXiv · 2026
Graph-Constrained Policy Learning for Extreme Clinical Code Prediction
Reframes extreme clinical coding as hierarchy-aware graph traversal and improves performance over flat prediction baselines.
Read papermedRxiv · 2026
FHIRTrustBench: A Benchmark for Interoperability-Driven Clinical AI Readiness and Trustworthiness
Introduces a multidimensional benchmark for evaluating FHIR-based clinical AI systems.
Read paperarXiv · 2025
Early Prediction of Multi-Label Care Escalation Triggers in the Intensive Care Unit Using Electronic Health Records
Predicts multiple forms of clinical deterioration using the first 24 hours of ICU data.
Read paperOpen research infrastructure
Software & Resources
Public tools and repositories supporting reproducible biomedical informatics research.
Clinical AI Workspace
MediLoom
A recent public lab repository supporting healthcare and biomedical data workflows.
View on GitHubNarrative NLP
BurnoutSurveillanceAI
Open research software for clinician burnout surveillance using narrative-based NLP and machine learning.
View on GitHubKnowledge Graph Authoring
KG4BioMedCntnAuthoring
Research software for enhancing biomedical semantic content authoring through knowledge representation.
View on GitHubNews & Updates
St. John’s Professor Awarded Grant to Study Use of AI in Medical Coding
St. John’s University has been awarded a $550,000 grant from the US National Science Foundation to develop an artificial intelligence (AI) remedy to the time-consuming process of medical coding and health care billing...
Read article
Dr. Bukhari was invited by the Florida International University to speak at KIFCKS Seminar Series.
Source: Dr. Bukhari was invited by the Florida International University to speak at KIFCKS Seminar Series.
Read article
Bukhari Lab participated at FLAIRS-35 Florida in their socio-technical approach to reducing the digital divide.
Source: Bukhari Lab participated at FLAIRS-35 Florida in their socio-technical approach to reducing the digital divide.
Read article