Satish Gupta currently works as Director of AI & Analytics at Cognizant. He provides global assistance for all R&D, Discovery, and Analytics for the company’s pharmaceutical, life sciences, and healthcare clients.
He supported the clinical and crop science applications of the Bayer Crop Sciences account as a life science domain consultant at TCS, Delhi. In addition, he was a team member that validated the NGS panels used in oncology to meet CAP/CLIA/NABL auditing bodies’ compliance requirements.
INDIAai interviewed Satish Gupta to get his perspective on AI.
It’s great to see someone with a bioscience degree employed in data science. How did it all begin?
Science is an evolutionary subject which keeps on upgrading through the implementation of new methods and technologies leveraged from research. Bioinformatics is one subject that gives life science students exposure to algorithms, databases, statistics and programming. All the aspiration for learning new topics and the demand for the application of bioinformatics into current scientific research has gradually pushed many of us into data science. There are many good universities and institutes offering bioinformatics courses and meeting the demand of the scientific and pharma sector. The application of third/fourth generation technologies into scientific research has poured vast amounts of data into our bucket to inspire us to learn more and make a meaningful interpretation of it. It is called the era of data and life science, healthcare and pharma sector has leveraged it quite beautifully.
Who motivated you to seek an AI career? What was the driving force?
I would say it was a gradual move, and “Bioinformatics” was a buzzword during our Master’s degree, and it touched us upon it. I was interested in starting my career in the industry after my Master’s degree in biotechnology, but I was not satisfied for many reasons. The hunt to join the industry made us realize the upcoming demand for bioinformatics. The bioinformatics course at JNU, New Delhi, gave me good exposure to databases, statistics and programming which motivated me to continue my work later in research institutes and pursue my career in the industry in different roles. There is a massive demand for resources in the modern way of looking at data. It is called “Explainable AI”, where these blends of expertise are well fitted. As soon as big data becomes a part of one’s journey, AI has to come along with it.
What were the earliest obstacles you encountered? How did you conquer them?
As mentioned, my current aim was to pursue a career in industry, but I needed help getting a break even after the bioinformatics degree was in hand. So, I started to work at the premier research institutes in India to gain experience to make a gateway into the industry, as they always prefer an experienced candidate over a fresher. I also connected with people working in academia and industry through various conferences, workshops and meetings. Proactive networking always works best for me. It also allows learning and being aware of new aspects in the scientific domain. After a couple of years working in a research institute, I got a break into the industry, but I soon realized the necessity of higher education for personal growth.
What are your responsibilities as the Director of AI and Analytics for Bioinformatics and Life Sciences at Cognizant?
It is quite a challenging role where I must keep updated with recent trends in the life sciences, healthcare and pharma industries. Cognizant is a service provider, and we, as a business unit, focus on the implication of AI & Analytics for our business partners based on the required objectives. Therefore, I need to understand the exact requirements from an R&D, discovery and analytics perspective and provide a solution strategy. At the same time, I also try to understand their broader theme of work and collaborations to gather pain points where we can support, provide a solution, and have a long-lasting business relationship.
Tell me about your PhD research. What were your research contributions?
The research work focused on investigating genetic and environmental modifiers on cancer risk. I was mainly involved in analyzing the modifying effects of selenium in the blood plasma/serum, and polymorphism in Selenium (Se) metabolizing genes on cancer risk in CHEK2 and unselected lung, laryngeal and colorectal cancer patients. I also explored the role of methylation in cancer-related and selenoprotein genes in breast carcinoma. Some of the conclusions were as follows:
- Higher Se concentration is significantly associated with a lower probability of cancer incidence.
- Se concentration can be a valuable marker for the early detection of cancers in the studied group.
- The effect of selenium level in blood serum on cancer incidence may depend upon genotypes in selenoprotein genes.
- Methylation of BRCA1 promoter in peripheral blood is associated with breast cancer risk in patients with BRCA1-negative germline mutations.
I also collaborated with multiple research groups and published >10 publications during my Ph.D.
Is programming expertise essential for bioscience graduates who want to work in artificial intelligence?
I highly recommend exposure to program language if one opts for a career in data science. It again depends on the demand of the role and responsibilities. For example, the data scientist would need more statistical knowledge with a fair understanding and experience in programming, and a data engineer, in addition, would also need an advanced level of algorithm development, experimental design & programming experience. Understanding cloud technologies is essential, as everything is deployed over the cloud. One can learn and upgrade their skills thanks to many online learning platforms.
What advice would you provide to someone who wants to work in artificial intelligence research? What should they focus on to advance?
AI is an application we can implement in different domains, from healthcare, banking, finance, market research, agriculture, climatology etc. Understanding any domain of interest and figuring out the challenges in that particular domain can be leveraged using AI. The following approach would be to look for available data and define a problem statement to solve using data science methods. Here I assume a prior experience with programming. Beginners can start by learning Python or R’s basics and data science modules. The flow I feel appropriate is a good understanding of the domain of interest, knowing at least one programming language, knowledge of statistics and cloud-based approaches, getting a good feel of data and implementing data science on the problem statement. There are many materials and courses on the web to get yourself certified.
What scholarly articles and publications have had the most impact on your life?
I have always worked on genetics, genomics and bioinformatics throughout my career. I admire articles, blogs and research papers about implementing AI/ML-based approaches to solve problems in the field of drug discovery and precision medications. It is interesting to read about the multi-omics process to analyze and interpret OMICS data, the integration of data from disparate sources and how we can implement FAIR guidelines. The post-COVID era has augmented the application of AI/ML approaches in clinical sciences. It is interesting to learn about decentralized trials and the extensive efforts to utilize Real World Data (RWD) for decision-making in patient recruitment, patient stratification and adverse drug reactions. AI has a significant role in the pharma industry, and the regulations of AI from the FDA and EMEA would be interesting to watch in the development of medical devices, hence shortening the duration of drug development.