Explore This Laboratory

Long Phi Le, MD, PhD

Director of Computational Pathology
Assistant Professor of Pathology, Harvard Medical School 
Assistant Pathologist, Massachusetts General Hospital


Computational Pathology is defined as “an approach to diagnosis that incorporates multiple sources of raw data; extracts biologically and clinically relevant information from those data; uses mathematical models . . . to generate diagnostic inferences and predictions; and presents that clinically actionable knowledge to customers” (Arch Pathol Lab Med 2014).

Other industries such as finance, e-commerce, social media and travel have benefited from access to and computation of structured, harmonized data to drive descriptive and predictive analytics. The same analytics and machine learning tools that have been developed for these industries could be leveraged to make our practice of pathology more effective, efficient and economical.

In the Center for Integrated Diagnostics, we have developed the computational pathology infrastructure to generate, capture and integrate genomics results with laboratory data. Having access to this integrated data store has greatly enhanced the practice of clinical genomics in the molecular diagnostics laboratory. By storing the data in a readily accessible database and combining it with a user interface for querying, pathologists, technicians, software engineers, bioinformaticians, data scientists, residents and fellows have been able to generate queries to explore the data for both clinical and research purposes. Interfaces have been built to take advantage of historical data to present descriptive analytics about variant detection across all prior cases. In addition, data scientists in the team have used  the data to generate several predictive models that are shown during clinical signout. These models include prediction of variant reporting, patient gender, sample swap, and microsatellite instability from the genomics data.

We have built a strong computational pathology team of software engineers, web developers and data scientists who will integrate the data that we generate across our pathology laboratories with the electronic medical record. The integration of pathology data with clinical data will allow us to explore, gain insight, derive hypotheses and generate models/tools to help with our day to day workflow. Our efforts will drive not only the clinical operation but also research and discovery.