Improving safety and access to immune effector cell therapy with artificial intelligence technology

PROJECT SUMMARY Immune effector cell-associated neurotoxicity syndrome (ICANS) is a cluster of symptoms associated with immunotherapy that did not show up in pre-clinical studies but has been found in 40-90% of the people undergoing immunotherapy as part of cancer treatment. The purpose of ICANS detection and prompt

treatment is to halt ICANS progression and minimize the risk of brain edema and herniation, the most feared sequelae of ICANS resulting in severe cognitive symptoms, coma, ICU stay, intubation, and potentially death. Current standard-of-care approaches to monitoring for ICANS consist of a brief neurocognitive assessment

resulting in an immune effector cell-associated encephalopathy (ICE) score indicative of the severity of cognitive impairment. However, recent work shows that there is an opportunity to detect ICANS at an earlier stage than is currently possible with ICE by monitoring for subtle early changes in speech and language such

as decreased fluency and coherence, word finding difficulties, and increased repetitiveness of speech. We have developed a Stress, Affect, Language and Speech Analysis (SALSA) system designed to administer and analyze speech-based neurocognitive tests over the telephone. SALSA conducts a conversation with the

patient and scores the audio received from the patient for several markers of speech fluency, verbal fluency and working memory deficits. Our long-term goal is to develop and validate an end-to-end AI-based solution for high intensity but low patient and provider burden neurocognitive screening for early manifestations of ICANS

based on the SALSA platform. The short-term objective of the proposed study is to examine the feasibility, safety and ability of SALSA to detect ICANS with specificity and sensitivity equal or better than standard of care in a prospective clinical study.