Skip to main content


Automated diagnoses from clinical test results

Trainee Achievements

Automated diagnoses from clinical test results

One way in which technologies that our IGERT trainees design can be utilized is to perform automated assessment of cognitive health. Clinical Psychology Alyssa Hulbert (an IGERT associate) and Computer Scientist Jennifer Williams (an IGERT trainee) have been working together to design machine learning techniques to automate diagnosis from clinical test results.

The trainees have found that using an extensive number (159) of tests , they are able to make classification with an accuracy above the published inter-rate reliability. Using only tests that the community has deemed clinically relevant, the accuracy drops below this level. However, using machine learning algorithms and automated feature selection techniques, our automated diagnosis is able to attain accuracy above the published inter-rater reliability. These results hold promise for automated diagnosis and informing the community about a cost-effective set of tests to consider for validation and use in clinical settings.