PathAI – Can we create a better, more robust pathologist

Can the transition to digital pathology, and associated use of AI to create a system that more accurately identifies disease than a human pathologist?

Industry Background: The Shift from Traditional to Digital Pathology

Traditional manual pathology is highly prone to subjectivity and variability from physician to physician. This can result in a totally different diagnosis for patients. Pathologists often spend countless hours poring over pathology samples and slides, making quick diagnostic determinations for patients. Pathologists are also in short supply, given the long grueling hours and lack of interaction with patients. More pathologists are retiring than the number of pathologists beginning their residency. Further, innovations in digital technologies allow for better image quality and processing.

This has created an environment where digital pathology has become a more mainstream option for diagnostics. Digital slides are created by taking images of glass slides with patient samples. These digital slides can be viewed, categorized, and manipulated on a mobile device or computer screen. Digital pathology, as a result, has several key benefits that make it more effective than traditional pathology.

About PathAI

PathAI is a Boston-based digital pathology startup founded in 2016. They layer on AI to enable substantial improvements to the accuracy of pathological diagnoses and the measurement of therapeutic efficacy for complex diseases. They have 2 major goals:

  • Help pathologists make more accurate diagnoses
  • Help develop new therapies by predicting how patients will respond to a pathologist’s recommended therapy

Helping Pathologists Make More Accurate Diagnoses

Through the use of AI applications, PathAI has been able to train their algorithms to analyze scans of pathology slides in similar way to human pathologists. They then deploy these algorithms alongside AI computational techniques to improve pathology diagnoses.

The organization has partnered closely with pathologists to develop data sets with thousands of slide images and annotations from pathologists to train their AI algorithms and develop the AISight Dx solution. For example, PathAI teamed up with Cleveland Clinic to use their AI-enabled pathology for translational research and clinical diagnostics. The five-year collaboration was inked to allow the “digitization of thousands of pathology specimens, creating whole slide images across multiple diseases.” As a result, they were able to significantly reduce variability and subjectivity between pathologists. Further, since the reading is done by a computer, the results are always consistent and reproducible.

Earlier this year, in August, AISight Dx received FDA 510(k) clearance as a solution for primary diagnosis in clinical settings. The web-based digital pathology slide viewing platform is interoperable with existing manual and digital pathology workflows. Due to this flexibility, the solution will be easy for pathologists to deploy with their existing workflows, allowing even “those who are working remotely to collaborate seamlessly to drive fast turnaround times for patients and improve patient outcomes.”

Developing New Therapies

More recently, PathAI has started to partner with major pharmaceutical and pharmaceutical services providers to deliver value to customers. PathAI can couple their digital pathology data with additional data (genomic information, EHR information, etc.) to better understand the patient’s scenario and develop additional predictive models.

Leveraging this concept PathAI developed AISight, which can be used by pharma companies, CROs (contract research organizations), academic institutions, and physicians to support AI-driven research.

Using AISight, translational researchers can use the digital pathology platform to better understand how patient tissue samples react to certain drugs and even predict how patients may react to new drug formulations. Further, AISight enables certain pharmaceutical companies to better understand the biomarkers (elements that predict disease) that react to a drug formulation. In certain cases, the solution can even be used to help companies develop new companion diagnostics, or diagnostic tests that can let you know if a patient will react to a certain drug.

PathAI has also partnered with major pharma services providers, such as Datavant. Datavant’s is a data aggregator and connector that helps use real world health data to improve patient outcomes. They enable companies to connect various datasets, allowing pharmaceutical companies to better understand how patients of various backgrounds react to certain drug formulations. PathAI also recently acquired Poplar Health, a laboratory services company which supports physicians across the US with its specialized testing labs. By combining the two companies, PathAI can further deliver an end-to-end service to customers, allowing for development and delivery of new AI-driven products for drug development.

Value Capture for PathAI

PathAI has spent the past 3-5 years inking major deals and research contracts with large players in the pharma and pharma services space (including LabCorp, BMS, Roche, Datavant, etc.). These deals are generally multi-year translational research deals worth millions of dollars, that enable these players to leverage the AISight platforms and the underlying AI to draw new conclusions that may speed up drug development timelines. Further, now that AISight Dx has received clearance in both the US and EU, PathAI can expect a steady revenue stream from hospital systems that decide to use the service to improve the accuracy and speed of their clinical diagnoses.

What’s Next

There are a tremendous number of options for PathAI, as they continue to build AI solutions for the pathology space. With time, as their AI engine collects more and more data, they can hope to eventually phase out pathologists, shifting them towards spending more time with patients or considering niche instances where the AISight Dx platform is unable to identify the correct course of action. Further, they have the potential to completely replace portions of the translational research pathway in the drug development lifecycle. They have the potential to become a major pharma services provider and completely redefine the role of the pathologist across care delivery.


C3 AI : Digital Transformation through Enterprise AI at scale


Gong: artificial intelligence for a non-artificial revenue lift

Student comments on PathAI – Can we create a better, more robust pathologist

  1. A very interesting journey indeed! Also intriguing to see the nature and sequence of their partnership deals to help them transition into a full-stack digital pathology as well as a pharma service provider. Also, their move to combining pathological data with other datasets for predicting patient conditions more accurately is really interesting as well. Thanks once again, for the wonderful blog!

  2. Thank you for the post, Aniruddh! I was really excited when PathAI first got the media’s attention several years ago. As you mentioned, its product can be used for improving drug development through more informed therapeutic use during clinical trials and post-approval and even for expanding gold-standard pathology services to regions that currently do not have access. Over time, it can be positioned to open up new opportunities within pathology and in other fields of medicine and healthcare. They can even move into radiology or genomics capabilities perhaps.

Leave a comment