Raxa-JSS Artificial Intelligence and Analytics

Patient-Facing Module

We are creating an interface for patients to our Health Information System that is lightweight and mobile, informative and responsive but most of all, gives agency and is empowering.  Our patients have a heterogenous mix of educational and economic capabilities as well as a large variation in educational and economic resources.  A large proportion of them are functionally illiterate (they cannot read or write their name) and many of the rest are not conversant with computer interfaces.  However, most of them have mobile phones that are becoming more "smart" every day which run on government-sponsored cellular data networks and they all have a growing appreciation for relevant information that is made accessible to them.  To enhance benefit delivery, huge government initiatives in India are now providing each patient with universal identification and exposing authentication services.  We will try very hard to use artificial intelligence methods to span this last but very important gap in bandwidth between device and human brain.  We believe that this effort is not only socially useful but also intellectually interesting and welcome ideas to take this forward.

A few ideas to start this conversation off are:-
1.  Patient Voice recognition of (in this case, Chattisgarhi) rural dialects connected with action engines (e.g. Siri, Google Voice + App Engine)
2.  Decision support and notifications.
3. Voice recognition -- identifies the speaker (e.g. using open source tools like CMU Sphinx)
4. Speech recognition --what is being said (in real time)
5.  --2 subclasses: small vocabulary (see IVR), large vocab-continuous
6. Voice dictation --recording speech
7. Voice transcription -- converting stored speech to text (DragocSpeak)
8. Text-to-speech -- vocalizing text (ubiquitous)
9. IVR (integrated voice response) -- "Please select from the following..."
10. DTMF (telephone touch tones) --using keypad to enter text/numbers
11. Predictive texting (?) -- autocompletion applied to keypad
12. Biometric identification  --fingerprint, retinal scan, voice recognition, facial recognition 
13. Animation -- in sync with text-to-speech or recorded voice for procedural info
Recorded voice --
Podcast --
Images --
geolocation -- you know
2-step authentication (right term?) -- I think it refers to both of:
  a) verifying through another channel (email for lost password)
  b) asking randomly selected questions from internal or public data

Provider-Facing Module