Member News

Leidos, Google Team Up to Develop AI-Powered Health Care Application

Domain-specific model

Leidos, Google Team Up to Develop AI-Powered Health Care Application

Leidos and Google are developing a health care application that uses a specialized generative artificial intelligence model to improve accuracy and efficiency in the medical field.

The application is built on Google’s Medical Pathways Language Model 2, which is trained on extensive medical data, Leidos said.

Srini Iyer, Leidos Health & Civil Sector’s senior vice president and chief technology officer, and Ken Su, Google Cloud AI for Healthcare Group’s senior outbound product manager, oversee the collaboration.

Iyer, a 4×24 member, emphasized the importance of using domain-specific AI models such as MedPaLM2 to interpret data correctly, as generic models may not understand the nuances of words and data in the medical context. Meanwhile, Su said the application also offers advantages as it provides responses that are more accurate and relevant than those from foundational models.

Iyer added that the collaboration yielded promising results, producing a demo version of a full-stack genAI application within six weeks.

Both parties also acknowledged the challenges of bias and security in using AI in health care. To mitigate risks, the application uses a retrieval augmentation generation tool to update itself with new data and employs human oversight to verify the model’s output.

According to Iyer, Leidos is looking to scale the application for future demands, including processing various data formats such as images and audio.

Potomac Officers Club Logo
Sign up for Potomac Officers Club's daily briefing
Receive updates on events and relevant news

Category: Member News