Senti - Wearable technology to enable remote Precision and Predictive medicine for respiratory patients.

Background: Senti is a novel wearable medical device: a smart garment which enables remote and autonomous chest examination.

Aims and Objectives: In parallel to seeking market approval for the Senti hardware, work will continue on developing machine learning algorithms to autonomously monitor chest sounds.

Our vision is precision and predictive medicine for respiratory patients, delivered through the creation of predictors of deterioration, assistance in diagnosis (e.g. differentiating exacerbations of heart-failure from COPD), personalising treatments (through acoustic phenotyping of heterogeneous COPD/Asthma), enabling patients to make more informed self-care choices (such as when to take rescue packs) and optimisation care pathways (identifying appropriate services for ongoing care).

Project plan and methods used: Our project runs in two streams. First; gaining market approval for our version-1 non-intelligent device. However; this grant will enable us to continue work in earnest on our autonomous system.

A dataset of respiratory sounds will be captured using our device, via the clinical investigation of our version-1 device. A number of models will be trained to classify chest sounds and their underlying diagnoses. A retrospective cross-validation approach will be employed to validate these models. Timelines for delivery: This proposal will be delivered over nine months, starting in September.

Anticipated Impact and Dissemination: Deploying the non-intelligent version of Senti (enabling remote chest examination) to community respiratory teams across the country will deliver immediate benefit to patients. However, the findings from our machine learning work will be published and disseminated.

These two streams will further impact by positioning us to undertake a more robust validation of our autonomous device through a prospective trial; whilst also enabling us to gather further data from patients using our non-intelligent device.

We will also be in a position then to complete a health economic analysis on the potential cost and benefit of our autonomous device.