About :
It's a mass diagnostics platform in which we can detect diseases even when people are coming in huge numbers. This kind of diagnostics platform is intended to reduce the work load on the medical practitioners and other medical workers who are under immense pressure due to the COVID-19 pandemic.
Why did I make it:
Recently my football teammate's father got diagnosed with COVID-19 and his condition was critical. After two days he also was tested positive. After 3 days he posted a video on twitter showing us how horrible the medical facilities here at Hyderabad, India were. When I was talking to him he also explained about the pathetic plight of the medical diagnostics system in the hospitals.
So that day I wanted to do something for him.
Working :
Unified Medical Monitoring InterfaceAim: To reduce the immense workload on the doctors and introduce cost effective automation in the medical diagnostics areas.
Problem: Due to the current COVID-19 scenario the hospitals and testing centers are getting flooded with patients. But even with the increasing number of patients our current medical technology fails to cope up with the needs of increasing patients as it offers a huge workload on the doctors, nurses and other medical personnel. The current doctor to patient ratio in India is about 1:1800.
But with the current conditions we need to have new technologies that can help us work with the limited medical personnel we have.
Solution: Building an AI based diagnostic platform that can give reports based on the data from X-Rays & other forms of scans.
Now this tool can help us assess the severity of the COVID-19 patient. Normally the patient’s chest X-rays are taken to help understand the magnitude of the viral infection as the level of pneumonia can be detected from the X-Rays. So, using NVIDIA’s Clara SDK we can build a software platform that can directly be hooked up to any medical viewer. This interface has the capability to connect to any machine like an MRI, CT, X-Ray, etc. Due to the current data constraint we are sticking to X-Rays. Once we have enough data, we can use transfer learning to get the medical application ready for the rest of the medical testing platforms.
Advantages:
Automates the diagnostics in the medical fields.
Highly accurate and simple to use
Fast report generation and cost effective
Disadvantages:
An AI can never be 100% accurate.
Scans have a batch limit of 100 scans per batch i.e. the AI can function at its highest accuracy if we limit 100 inputs per batch of X-Rays.
Phase - 1:
Outputs:
Video Presentation of our end to end solution:
Further Scope of development(Phase -2) :
-> Very soon we are planning to integrate various test platform like CT, MRI , etc and also integrate various disease detection CNN's into this unified platform so that we can harness its batch processing capabilities and deploy a highly salable and an efficient model that will have the potential to become one of the medical diagnostics revolutions.
Ending Note :
I have also designed an ESP32 board that can special features like 40 volts regulation , level shifted i2c etc. This board can be used to make IOT based patient monitors and ventilators . As I believe in open source I have attached the complete Gerber file so that designers can use it directly. Would be truly glad if its of any use.
Comments