An advanced clinical records and productivity assistant

TwynSyght is a smart, domain expert built application that extracts summary information from electronic mental health records. It uses machine learning to pull decision specific details from electronic health records saving clinician's and other user's time. Time which can be spent on patient care and decision making. It increases safety because key information from an entire record is made available to professionals who don't have the time to read the entire record. It is aimed at all levels in a learning mental health system from front-line clinicians to service managers, regulators and investigators.

Official IBM Watson AI XPRIZE Team

TwynBrain has entered TwynSyght into the IBM Watson AI XPRIZE. This is a prestigious $5 Million competition now in it’s 3rd round. TwynBrain is one of 34 teams in 2019 reduced from a field of 148 in 2017 and is also the only remaining UK team.

The IBM Watson AI XPRIZE time line has a rigorous judging process with completion of our ethical framework in May 2019 and semi final selection in October 2019.

Security

  • TwynSyght is a cloud based application. It runs summaries in real-time, so it does not store patient data.
  • TwynSyght encrypts all patient data sent to it’s servers for processing.
  • Multi-factor authentication protects our customer accounts.
  • The TwynSyght machine learning models are already fully trained on the mental health domain. Patient data is therefore not needed for any training.

How good is TwynSyght at extracting information?

TwynSyght currently extracts data in the following domains with the associated performance metrics:

Summary Type Recall (sensitivity) Accuracy Matthews Correlation Coefficient
Mental State 0.83 0.83 0.82
Care Planning 0.89 0.89 0.88
Diagnosis 0.92 0.92 0.91
Life Events 0.86 0.86 0.84
Mental Health Risks 0.96 0.96 0.95
Severity 0.95 0.95 0.89
Treatment Methods 0.98 0.98 0.98
Average 0.91 0.91 0.90

Instinctive ease of use.

TwynSyght is intended to be very easy to use. The demo clip below shows the selection of a sample record, selection of life events and summarisation with visualisation using default options.