Role and Learnings

AiOmix is a UAE based startup implementing AI pipelines on multi-omics data to pre-diagnose prevalent human diseases.

Here, my role was quite diverse spanning from core business to Product Development. I was involved in analyzing the market potential (in terms of region and customer segment), competitors and global opportunities. At product side, I helped in identifying, collecting and preparing data for AI model training and testing purposes. I also played an important role in model improvement and final validation.

While working there, I learnt:

the thought process of how AI works
concepts of data analysis: Data Collection and Preparation
working in a cross-functional team (all other were from CS and Data background)

Work and Impact

Market Trend and Competitive Analysis:

As a startup, we wanted to overall trend and opportunities available in the market so, I performed Market and Competitive analysis. I did secondary research (company websites, articles, govt data, WHO data etc.) and primary research (interaction with industry professional and academic professors) to perform below work:

went through industry articles on new developments and technology
identified healthcare market opportunities and prevalent diseases: world wide and region wise
created a inhouse database of all global competitors focusing their specialty and operation region

Impact:
Based on the research, we decided to work on cardiovascular disease as it has more opportunity in Middle East region.
We also planned to use Computer Vision in Cancer diagnosis using chromosomal and genomic data.

Data Preparation and Product Validation (AI pipelines):

We built two AI pipelines: Named Entity Recognition (NER) and Relation Extraction (RE) pipelines. To train these pipelines we needed good quality data.

Being an bioinformatician, I identified the required data as an Subject Matter Expert; made data ready for pipelines: Collected, Cleaned and tagged.
For data, I used 40+ public biological data repository, mainly, Cancer Genome Atlas, NCBI, DB etc. For cleaning and tagging purposes, I used Inception (AI service) and Notepad.

Impact:

Team achieved F1 Score of 85% and 80% respectively for NER and RE pipelines.