Abstract
Suicidal ideation is a major health concern in the United States, with many millions of people reporting experiencing serious suicidal thoughts each year. Early detection of suicidal thought is critical in preventing suicide attempts and treating affected individuals. Recent research has shown how machine learning can be used to detect suicidal ideation from phone speech data. However, given the very sensitive nature of the data involved in this process (i.e. phone conversations of at-risk persons and prediction results), it is difficult to imagine how such an application could be used in practice. To address this issue, we investigate a privacy-preserving variant of the ideation detection application flow involving homomorphic evaluation of neural networks. We describe multiple realistic use-cases to aid both affected individuals and clinical practitioners that would be enabled as a result of this secure infrastructure. We also give first order performance estimates for homomorphic evaluation of the networks proposed, and discuss various opportunities for further analysis.
Original language | English |
---|---|
Title of host publication | Protecting privacy through homomorphic encryption |
Editors | Kristin Lauter, Wei Dai, Kim Laine |
Publisher | Springer Cham |
Pages | 133-146 |
ISBN (Electronic) | 9783030772871 |
ISBN (Print) | 9783030772864, 9783030772895 |
DOIs | |
Publication status | Published - 05 Jan 2022 |