NIH Award Funds A.I. to Detect Alzheimer’s in Clinical Notes

(HypnoArt, Pixabay)

17 Feb. 2023. A small-business grant to a developer of natural language analytics for health care aims to detect evidence of cognitive decline from physicians’ notes in electronic health records. Melax Technologies Inc. in Houston is partnering on the project with researchers at Brigham and Women’s Hospital in Boston, funded by a two-year $2.5 million award from National Institute on Aging, part of National Institutes of Health.

Melax Technologies is a six year-old company developing software using natural language processing, a type of artificial intelligence, applied to health care. The company says its core software package, known as clinical language annotation, modeling and processing toolkit or CLAMP, is built on algorithms trained by large volumes of records and documents to recognize and automatically encode clinical information. Melax Tech offers CLAMP as a stand-alone package for use on a client’s own systems as well as via the cloud.

The National Institute on Aging award funds a system designed to detect the earliest stages of cognitive decline leading to Alzheimer’s disease and related types of dementia. For many patients, early detection of cognitive decline is done by primary care physicians, who may or may not have the training needed for this task. Melax Tech proposes creation of deep learning algorithms with medical informatics experts at Brigham and Women’s Hospital, affiliated with Harvard Medical School, for detection of cognitive decline in notes made by primary care physicians. The company plans to build on an earlier collaboration with the hospital resulting in an initial deep learning algorithm for cognitive decline detection based on physicians’ clinical notes.

Natural language processing and deep learning algorithms

In the new project, Melax Tech and Brigham and Women’s Hospital expect to further develop and implement new natural language processing or NLP and deep learning algorithms in a decision-support system for physicians, and validate the system in a primary care setting. The project team plans to work directly with physicians in developing the decision-support tools, to help them better identify patients at risk of cognitive decline and provide personalized recommendations for care.

“We will develop cutting-edge NLP algorithms,” says Melax Tech NLP director Jingcheng Du in a company statement, “to identify at-risk patients and extract cognitive concerns, symptoms, diagnosis, assessments, and social determinants of health factors from clinical notes.” Du is the company’s principle investigator on the project.

The researchers expect to validate the clinical decision-support tools with primary care physicians and real patient data stored in electronic health records at Mass. General and Brigham and Women’s hospitals that share the same system. If the system is successful, says the team, it will be the first automated clinical toolkit for early detection of cognitive decline validated by physicians.

The award is a Small Business Innovation Research or SBIR grant made under NIH’s small business programs that set aside a part of the agency’s research funding for U.S.-based and owned companies. Most SBIR grants are made in two parts: a first phase to determine technical and commercial feasibility, and a second phase to develop and test a working prototype or prepare for clinical trials. This is a second-phase project, since the earlier Melax Tech/Brigham and Women’s work serves as the equivalent of a first phase.

More from Science & Enterprise:

• Quantum, AI Systems Company Gains $500M in New Funds
• AI-Aided Mobile App Shown to Screen for Melanoma
• Start-Up Creating Software to Simplify Biomedical Analytics
• Real-World Data Shown to Match Eye Disease Trial Results
• Audio Digital Biomarkers Start-Up Raises $3.7M in Seed Funds

We designed Science & Enterprise for busy readers including investors, researchers, entrepreneurs, and students. Except for a narrow cookies and privacy strip for first-time visitors, we have no pop-ups blocking the entire page, nor distracting animated GIF graphics. If you want to subscribe for daily email alerts, you can do that here, or find the link in the upper left-hand corner of the desktop page. The site is free, with no paywall. But, of course, donations are gratefully accepted.

*     *     *