Peace

APGAR Scoring

Inspiration

In our country, the mortality rate for newborns is alarmingly high. This is mainly due to the lack of adequate attention to newborns when they are in their most vulnerable stages - during the first few hours. A very important procedure that is carried out during this time window is "APGAR scoring", wherein the general health of a newborn is judged based on 5 criteria: Appearance, Pulse, Grimace, Activity, Respiration. Unfortunately, even in the 21st century, APGAR scoring is completely a manual process. This score is very crucial as a low score can indicate that the baby requires immediate medical attention. This inspired us to develop a smart and automated solution.

Thanks to Mr. Rajendra K and Rakshith

What it does

APGAR is an acronym for: Appearance, Pulse, Grimace, Activity, Respiration. APGAR scoring is based on a scale of 0 to 2 for every aspect, and an overall scale of 10. Our idea is to automate as much of the scoring mechanism as we can.
The Python algorithm analyzes and classifies the baby's conditions
This data is pushed to the Android client (via GCM)
Medical professionals can use the Android client to stay updated

Scope of the project includes: Appearance, Grimace, Activity.

Monitoring the appearance of the newborn is accomplished
Classifying the facial expression of the newborn
Determining activity levels of the newborn

How we built it

OpenCV, SkImage and DLib libraries for image processing
Android Studio to develop the Android app

Challenges we ran into

Obtaining facial images dataset for newborns

Accomplishments that we're proud of

Novel idea, first automated solution for APGAR scoring.

What we learned

Video processing
Feature extraction
Structural Similarity Index to determine activity levels

What's next for Connected APGAR

Automate the entire scoring mechanism, including Pulse and Respiration
Implement an alert mechanism to notify medical professionals if the baby requires immediate care