TRIAGE Telephone

Searching for buzzwords in series of text
Flow chart showing mental health diagnosis steps with TRIAGE telephone
Signal processing of data, showing band pass filter, letting certain frequencies through and remove the rest

Inspiration

The inspiration was from an experience of mine at the SAMHS (Student Access to Mental Heath). After arriving at the clinic, I was asked to complete a questionnaire, but I felt rushed and was not comfortable, since the questionnaire was unexpected and I did not want to be late for my appointment. Furthermore, I felt like I did not express my problems well to the counsellor in person. My overall experiences could be improve if I was given more time to do the questionnaires, as well as knowing what to expect in the appointment.

This project aim to aid clinician in diagnosing mental health issues, offer different medium of communication to patients and lastly IAPT (Improving Access to Psychological Therapies) dataset and Patient Reported Outcome Measures (PROMs) by collecting and analysing voice data.

What it does

The TRIAGE Telephone is a phone bot (Dr Dan) which call patients prior to their appointment to run through a series of questions. The patients will first answer a general open ended questions to state their problems, then they will be ask series of questions based of the PHQ-9 (Patient Health Questionnaire), which they will answer by pushing the keypads.

We planned to implement further data analysis based on the open ended questions phase. We would take the patient's recorded voice and turn it into text through IBM Watson. This text would then be analysed for certain "buzzwords", for example "sad","hopeless" etc. The answer also analyse for the most common words uses, except stopwords such as and, the. The voice recording is then analysed in terms of pitch and frequency to detect symptoms of anxiety and depression, such as shaky voice, quiet voice, stuttering, inability to communicate concepts and hesitancy.

Once the data is collected, it would be sent via encrypted email prior to the appointment, where they can analyse it in their own time and pace. This will save time for the patients, the clinicians, improving work flow and provide a more accessible way to analyse and diagnose mental health issues.

How I built it

We built this application using Node.js , Nexmo's Call API to get user input, and nodemailer for sending the results to the doctor's email.

Challenges I ran into

As two of the team had never used node.js before, there was an initial learning curve in becoming acquainted with the language. In addition, we encountered issues implementing more complicated data analysis tools and therefore had to scale down the scope of the project.

Accomplishments that I'm proud of

Learning node.js and implementing nexmo's APIs. Implement the buzzwords search on Moby Dick

What I learned

Node.js.

What's next for TRIAGE Trio

Implementing the more complicated data analysis, data permanence via some kind of database and persistent user accounts. We hope to expand this service to other TRIAGE assessment, such as for emergency services

We are applying to all the challenges