My first research project

Unsuccessful story of THERMODOCS

---

Hey readers!

I’m a first year student @IISER TVM and, This blog is about my first research story.
Now as we know for any project we first give a hypothesis and then do experiments, research and other things to check whether it’s true or false.
My hypothesis was “Diagnosis using fever pattern i.e, Temp vs Time graph”

So, the idea was to take the advantage of the fact that different disease have different temp. vs time curves recorded for a same period of time.

For example

In case of malaria; Temperature rises sharply and returns to normal, often at regular interval.

In case of Dangue; Fever subsides temporarily and returns after 2-3 days.

In case of Borrelia infection; Periods of high fever alternate with periods of normal temperatures.

So yes, temperature vs time graph can provide valuable clues to predict the type of fever of disease a person is suffering. I went through a lot of research papers, hospital reports, and had conversations with some doctors. Ultimately I got to know that this method has been used traditionally but never digitalised. YES, my main aim was to digitalise this method via a software so as to make it access to places which lack quick or better or emergency medical services and also to people where services are there but not frequently affordable for them.This project also targeted the large population of india which directly go to pharma shops instead of a real doctor.

I started working on my AI model which would take inputs,process it,match it with existing graphical data and provide diagnosis.

BUT,unfortunately I found that this method is NOT DEFINITIVE.

So what does this ‘not definitive’ exactly mean. It means that there are sometimes overlap between fever pattern of two different diseases.For example Malaria and sepsis had similar graphs in some cases which I tested.

Now as you guessed,it was time to either end my project and let the Null hypothesis win or do some more brainstorming and find a tool or an idea to enhance the success rate of my AI software.

Yup, I went for second option and tried many different methods, tools, tests and the result was ….
…...

YUP, I did find a solution to this issue. The solution was simple. It was…..(Ooops, I’m out of word limit for this blog.Well let’s continue the story in next blog.)