Opendata, web and dolomites

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - Limbic (An API for Emotional Intelligence)

Teaser

Limbic is an API platform that senses human emotional state using signals from the user: facial expression, voice, text, and importantly, physiological signals. Our software developer kit (SDK) is one line of code that sits inside client apps. Our SDK detects which data...

Summary

Limbic is an API platform that senses human emotional state using signals from the user: facial expression, voice, text, and importantly, physiological signals. Our software developer kit (SDK) is one line of code that sits inside client apps. Our SDK detects which data sources are available (camera, microphone, text corpus, and wearable heart sensor) on top of supporting information such as age, weight, and gender. It then sends this to our servers where proprietary machine learning algorithms return predictions on the user’s emotional state. Our machine learning system represents a new brand of ‘Emotion AI’. Limbic is guided by a broader definition of intelligence than present in AI today. In the near term, our software will augment digital products, providing developers with new ways to interact with their users.

Following a successful application to Phase 1 of the EC Horizon 2020 SME Instrument, we carried out a feasibility study for our innovative technology and we have defined a Minimum Viable Product for development. Based on Limbic’s proprietary physiology-based emotion AI, we plan to build a scalable solution for mental health monitoring.

Definition of market problem:
Depression and anxiety costs the global economy $1 trillion annually. 1 in 4 people in Europe experienced a mental health issue last year, and almost 60,000 people in the EU will take their own life this year. However, unlike other areas of medicine, mental health lacks quantifiable metrics. This makes treatment less efficient, and makes it challenging to evaluate treatment response.

Product development plan:
Following this Phase 1 feasibility study, Limbic plan to build a bespoke product for patient mental health reporting. This will include three parts: 1) a fitness tracker, 2) a patient mobile app, and 3) a clinician web app.
1) Through the fitness tracker, patient heart data will be streamed to cloud servers where Limbic’s algorithms will predict emotional state.
2) When a significant emotional event is detected, the patient mobile app will prompt the patient to answer therapeutic questions: “How are you feeling? Where are you? Who are you with?”
3) By proactively engaging with the patient in real time, during relevant moments, Limbic will relay valuable patient data to the clinician via a web app.

Work performed

Limbic has achieved a number of key results critical for the business during this feasibility study. These include:

- Pilot with a meditation app
Designs on how Limbic API would be used to build new features in a The Mindfulness App. Discussions with their engineers on how best to package Limbic’s proprietary algorithms for use by third party apps. No Go decision on this part of the business.

- Pilot with Garmin
Scope technical issues for working with wearables. They gave us 6 devices to use.
Had 5 meetings in total. Attendance from Global Product Lead for Garmin Health, 2 Product Architects, and a team of 4 engineers.

- Pilot with Polar
Scope technical issues for working with wearables. They gave us 10 devices to use.
It was helpful to benchmark against Garmin.
Had 4 meetings in total. Attendance from Research & Technology Director and Senior Researcher.

- Pilot with Oura Ring
Scope technical issues for working with smart rings. Understand requirements of new ring wearables market.
Gave us 10 devices to use.
Face-to-face meeting with Co-founder and Chief Innovation Officer, Chief Scientist and Sales Director.

- Meetings with Motiv Ring and Fitbit
Numerous exploratory meetings and NDAs signed. However, no formal pilot came from these.

- Clinician interviews
Interviewed 35 cognitive behavioural therapists. Design spec. for clinician dashboard.

- Patient interviews
Interviewed 50 patients undergoing cognitive behavioural therapy for depression/anxiety. Design spec. for patient app.

- Establish a feasible product development plan.
Set up a detailed product development plan, supported by the agile development methodology, to list the technical activities to be performed.

Final results

\"Limbic are developing the world\'s first wearable emotion-detection AI for mental-health patients to:

1. Enable more efficient use of clinician time within cognitive behavioural therapy (CBT) consultation - 50% of therapist time is spent getting answers to basic questions that could be achieved outside of consultation.
2. Reduce patient wait times.
3. Permit a more data-driven approach to evaluating treatment response.
4. Provide deeper insight into patient mental state, reducing the risk of premature discharge.

Limbic\'s solution uses novel AI, compatible with consumer fitness-trackers, to continuously/passively monitor specific measurements of cardiac functioning to predict patient emotional state. Upon detecting a significant \"\"emotional event\"\", a patient-facing app prompts the user to answer a series of questions that a therapist would ask if they were present. By collating-analysing-displaying user-input data, healthcare professionals can unobtrusively track their patient\'s mental state throughout therapy.

Potential impacts include:

- By monitoring panic attacks and emotional state passively through the wearable, Limbic can prompt patients to answer specific questions at relevant moments whilst on the waiting list to see a clinical psychologist. Limbic can then fast-track the assessment process by providing all necessary information to the psychologist in the form of an easy-to-view dashboard. Limbic will also save time in therapy, where clinicians spend ~50% of their time gathering basic patient symptom information. This could cut 8-10 sessions down to 5-7, and generate significant savings to healthcare services.
- ~30% fewer clinic visits for patients as the app will reduce the number of necessary check-ups, meaning less time spent out of work and less travel for patients.
- Greater levels of access to therapeutic support (digital tool is easier to distribute and use on daily basis), 2/3 people with known mental health disorders don\'t receive treatment (WHO).
- Reduced relapse rate and early treatment for people who have been discharged from healthcare services.
- Save patients, family and carers unnecessary travel and associated cost. Improved access to therapies will help tackle the 70million working days lost annually due to mental health problems.\"

Website & more info

More info: http://www.limbic.ai.