Data science professionals should have domain knowledge: Sreetama


From researcher in computational biology and structural biology / biophysics to senior data scientist, Sreetama Das has acquired an extensive knowledge base and has worked in various industries such as manufacturing and healthcare. Analytics India Magazine met Das, who is currently a senior AI ML engineer at GSK, to understand his ideas on AI-based solutions in digital health.

OBJECTIVE: Given your experience in a broad career spanning data science to AI and machine learning, what do you think the future of the tech-medical space will be?

Sreetama Das: To briefly mention my background, I worked with data science and machine learning applications for biomolecules as part of academic research during my PhD and then worked in various industries like manufacturing and health. This experience shaped my vision of what I think the field will evolve.

Data & Analytics Conclave>>

Several areas of the tech-medical space use machine learning for better results – for example, research and development for drug discovery or reuse, clinical trial optimization, manufacturing and quality control, digital health sensors, digital triage of pathology patients, to name a few. The data can be digital, well structured in tables, or huge images or even messy text – this is what makes problem reporting in healthcare exciting but difficult. Advances in computer vision and natural language processing will improve solutions or even solve problems that were not previously possible. In addition, solutions focused on explainability will be better accepted. The general trend is towards faster development, efficient manufacturing, better quality control, and easier access to health monitoring, disease detection and treatment outcomes for patients. In short, AI-based solutions will help improve lives.

OBJECTIVE: There have undoubtedly been significant advances in digital health. Was there a specific situation that prompted you to do additional research?

Sreetama Das: Digital health has seen significant advances in the development of digital sensors for health monitoring and disease detection. I will cite an example from a project I participated in in my previous organization (Engineering and Business Solutions Robert Bosch, India) since I have only been with GlaxoSmithKline for a short time. There, the team developed a new non-invasive hemoglobin monitoring sensor based on photoplethysmography. We collected data from many participants, both healthy and fragile, and trained machine learning models. As a result, our device performed better than some existing non-invasive hemoglobinometers, which tended to overestimate hemoglobin levels. The project required a lot of research and our results have been published in several reputable scientific journals.

GOAL: Enterprises have started to migrate data centers and on-premises applications to cloud and SaaS solutions. How feasible is this transition to the cloud given the many security issues and ransomware attacks that are occurring?

Sreetama Das: I’m not an expert in this area, but I am familiar with commercial private cloud and hybrid cloud offerings that combine the benefits of the cloud with the security of on-premises solutions and are used to address regulatory concerns. Additionally, we are starting to see the use of blockchain in commercial cloud solutions and upcoming technologies such as federated learning and its use in medical technologies. So, I think the transition is doable with some thoughtful strategies.

OBJECTIVE: In what we call a “combination of AI and digital health”, what are the near-revolutionary developments that also have future potential?

Sreetama Das: One of the revolutionary developments of recent times would be the development of AlphaFold (by Deepmind) which revolutionized the science of protein structure solution and has important implications for drug design. The software provided nearly accurate theoretical models of the 3D structures of proteins, which are currently obtained by time-consuming and expensive experiments or sometimes represented by not-so-precise theoretical models. The implementation of the models generated by AlphaFold will significantly reduce the search space, thereby reducing the time and expense of the drug discovery process.

Other advancements include deep learning in several areas, for example, in classifying chest x-rays of normal influenza versus COVID-19 infected cases with high accuracy or in early detection based on respiratory sounds. abnormal voice. In addition, image and video analysis finds applications in monitoring physical activity or elderly patients in the event of falls or other problems. Finally, personalized precision medicine is also an area with great potential.

OBJECTIVE: In general, why do you think there are widespread misconceptions about artificial intelligence in healthcare, and why do they persist?

Sreetama Das: I think there are several sides to this question. First, when there are breakthroughs in artificial intelligence algorithms, many articles in general forums are often very flattering, without trying to explain how they work, the input requirements, or where they can fail. As a result, there is a disappointment when such tools fail to generate the expected results, either due to improper application or improper data. It is important to remember that collecting enough good data remains a challenge for many health issues. In addition, some problem statements may not be feasible and require modification after several rounds of discussions with stakeholders. In addition, health care is a sensitive topic, and people are often (and sometimes rightly) fearful of accepting “black box” solutions or the “fairness” of such solutions.

Therefore, raising awareness of artificial intelligence and solving problems – data requirements, understanding how machine learning works and what is feasible, and explainability of models – will help eliminate persistent misconceptions.

OBJECTIVE: Could artificial intelligence possibly replace clinicians?

Sreetama Das: Human biology is very complex – we still don’t understand everything. Clinicians make decisions by looking at many different aspects and based on their experience. It is difficult to design “general” machine learning solutions that can work the same way as a skilled clinician – only solutions to very specific AI can develop well-defined problems for which a lot of data is available. Since healthcare delivery has a direct impact on people’s lives, the future would have AI solutions “assisting” clinicians in their decision-making rather than completely replacing clinicians.

See also

OBJECTIVE: What resources, such as books and journals, would you recommend to aspiring professionals in this field?

Sreetama Das: There is often a lot of useful information in blogs on Medium. I refer to GitHub for code bases. Other sites to look for will be articles with code and corporate blogs for recent developments (eg Google, Facebook, Microsoft). There are also many free videos and course materials (e.g. MIT Courseware) for future professionals.

OBJECTIVE: What would you recommend as an effective career direction for professionals who want to succeed in the digital health space?

Sreetama Das: I will answer this question based on my background. Data science professionals need to have some domain knowledge to develop appropriate solutions. Since digital health is a vast space, it is important to be aware of it and to identify the topics that interest and understand professionals.

It also helps to be flexible and adaptable to change, as different areas can gain traction at different points in a professional’s career. Finally, we must be lifelong learners and acquire relevant knowledge when the need arises.

Finally, it is important to focus on the foundations of such careers. Good coding skills, an understanding of key concepts and good communication are important. Also, keep in mind that such projects are often based on teamwork, with different skills within the data science team (analyst, ML engineering, MLOps, etc.). It is therefore important to talk with your peers and learn more about these aspects.


Subscribe to our newsletter

Receive the latest updates and relevant offers by sharing your email.


Join our Telegram Group. Be part of an engaging community


Source link

Comments are closed.