Computational Healthcare

De-identified data on millions of patients is available for research

Computational Healthcare platform simplifies use of this data while preserving privacy

  • Analyze 150 Million visits

  • Conduct more than million studies

  • Enforce privacy policies on computations

  • Investigate future visits by similar patients

  • Train Machine Learning models

  • Evaluate model performance

  • Discover novel associations

  • Conduct group studies with physicians

  • More information coming soon!

We strongly believe in reproducibility. We provide code to replicate results and for comparison with approaches such as differential privacy.
Computational Healthcare library

Please visit this website on larger screen (laptop/tablet) to view demos.

Aggregation Framework

DUA Compliance

Guarantees DUA compliance by enforcing policies at computation.


Aggregated statistical reports can be quickly visualized. The standardization allows reuse of charts, widgets and tables.


Extend framework

The aggregation framework can be extended to incorporate additional data elements.

Generate Reports

Save aggregated statistical reports in Excel or CSV format.

Manage & Index Data

Computational Healthcare indexes and aggregates healthcare data from millions of healthcare visits. The user interface simplifies data management, analysis, replication and dissemination.

Aggregation Framework

The novel Aggregation Framework allows quick analysis and exploration at scale, while maintaining confidentiality and privacy. It also automatically ensures compliance with Data Use Agreements.

Aggregation Strategies

Analyze utilization and timing of inpatient procedures. Track patients over multiple years. New strategies can be developed which can go beyond simple combinations of diagnosis and procedure codes.

Search Engine

The aggregate statistics can be easily searched using keywords and codes. Users with access to underlying data can quickly select and perform additional studies. Can be used for secure data delivery in future.

Inpatient Procedure analytics

Analyze timing and utilization of procedures during hospitalization

  • When do patients with leukemia receive transfusion?

  • How common is ECMO followed by thoracentesis?

  • How many patients with Sepsis were Admitted via ED and then Transferred to another hospital?

Readmission & Revisit Analytics

Track rehospitalization by specific reasons and All Cause Rehospitalizations.

  • Compare subsequent visits with initial and all discharges.

  • Track patients over time periods ranging from 7 days up to a year from discharge.

  • Track complex transition patterns such as Treat and Release ED visit following discharge or Ambulatory Surgery visit followed by Inpatient Hospitalization

Patient Analytics

Examine multi-year outcomes of disease episodes.

  • Do patients with MS suffer from more falls and injuries?

  • What are rare co-morbid conditions in patients suffering from diabetes?

  • How many times does a patient with poorly controlled diabetes visits an ER a year befor undergoing amputation.

Predictive Analytics

Develop, Train, Evaluate and Visualize prediction models.

  • Predict short term (Rehospitalization, Revisit) and long term outcomes.

  • Derive complex features. Train models using TensorFlow.

  • Evaluate and visualize performance of models on custom patient populations selected using other criteria.

Authors & Contact

Akshay Bhat

Akshay is a PhD candidate in Information Science at Cornell Tech (in New York City), Cornell University. Computational Healthcare is part of his PhD Research.

Peter Fleischut, MD

Peter is Chief Innovation Officer at New York Presbyterian Hospital and Associate Professor of Anesthesiology at Weill Cornell Medical College, Cornell University.

Ramin Zabih

Ramin is Professor of Computer Science at Cornell Tech (in New York City) and Professor of Radiology at Weill Cornell Medical College, Cornell University.

The information contained on this site is intended for educational purposes only. It does not constitute medical advice, nor is it a substitute for medical advice. You should consult a physician regarding medical diagnosis or treatment.

This study is not endorsed by Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality. All results shown on this site are derived from synthetic data.

© 2015 Akshay U Bhat, Peter M. Fleischut & Ramin Zabih, Cornell University.
All Rights Reserved, At this time we are pursuing the patent process to protect this software.