A diagnostic, prognostic, and patient monitoring algorithm that compares text and speech characteristics to a corpus to distinguish patients; validated in suicidal vs. non-suicidal adolescents and epilepsy surgery prediction.

This technology uses an algorithm to analyze text and speech by comparing words to a database of semantic and episodic concepts, which includes a range of similar concepts to ensure full understanding of the words. It can accurately understand the mental or cognitive state of an individual by comparing against a known corpus. This algorithm can be further combined with speech and voice characteristics to improve accuracy of patient categorization. A pilot study has shown the ability to identify speech of suicidal and non-suicidal adolescents with a high degree of accuracy, and it is being further validated via an app in Cincinnati Public Schools. In addition, it has also been validated with clinical notes of epilepsy patients to predict which patients are candidates for neurosurgery. Information-theoretic and machine learning techniques are used to determine whether and how sets of clinical notes from patients with intractable and non-intractable epilepsy were different. With additional corpi, this technology could be used to diagnose, prognosticate, or monitor patients' mental and cognitive states. It can provide healthcare workers with objective measures to improve clinical assessments.

Improved accuracy over other NLP techniques; potential for combination with genetic factors and facial features; integrates with EMRs; and allows for faster prognosis and treatment

More than 3M people in the US have some form of epilepsy. An estimated 11 attempts occur per every suicide death.