DATA STORIES | DIGITAL HEALTHCARE | KNIME ANALYTICS PLATFORM
Closing the Gaps in Healthcare with Low-Code
My Data Guest — An Interview with Chenny Solaiyappan
Co-interviewer: Alexandra Quintana
It was my pleasure to interview Chenny Solaiyappan in this new episode of My Data Guest. Chenny is an active and expert data scientist and KNIME user in the area of health insurance.
He uses KNIME to acquire, transform, and visualize data for health insurance businesses to consume. He currently works for Cognizant in Ann Arbour, Michigan, as a Director of Product Management & Strategy for TriZetto Government and Quality solutions. Our guest talked about the opportunity of making healthcare data and insights more accessible and effective by using no-code/low-code machine learning tools like KNIME, the pivotal role of data analytics in healthcare, and the significant impact of data-driven decision-making on healthcare quality and patient outcomes.
Disclaimer. The interviewee has expressly requested clarification that the opinions and perspectives shared in this interview are entirely his own and should not be taken as representative of any specific organization or entity.
Rosaria: Hi Chenny, I mentioned that you work in the health insurance business. What do you do exactly? And how does data science fit with your job?
Chenny: I work for an organization called Cognizant where our mission is to engineer modern business to improve everyday lives. Within Cognizant, we have a product house called “TriZetto Healthcare Products” that today touches around 200 million lives in the US. Within that segment, our customers are healthcare insurance organizations.
Healthcare is a heavily regulated industry, so we created within “TriZetto Product House” a “TriZetto Government and Quality Solution”. When governments want to invest in healthcare, for example for elderly or underserved population, they want to make sure the taxpayers money is well spent so our job is to measure the quality of healthcare. We are talking about public health, the science of protecting and improving the health of people and their communities using 90 different quality measures, so ours is definitely a data centered business.
Rosaria: Can you tell us about some typical use cases in health insurances, so that we can get a clearer idea?
Chenny: Consider that every year the US spends 4.3 trillion dollars on health care alone out of a total GDP of roughly 20 trillion dollars. It is a huge sector, therefore our objectives cover different areas:
- Maximizing wellness of patients. For example, by organizing preventive screenings.
- Improving healthcare quality and patient safety.
- Reducing costs.
- Finding a solution to the current labor shortages. We have a strong need for professionals with advanced skills in machine learning but it’s often hard to find those people. The low code, no code approach is literally addressing this issue by removing barriers for domain experts and other data professionals to build predictive models.
- Addressing physicians’ burnout. Doctors are inundated with so many regulatory challenges and we help them navigate this situation.
- Changing consumer behavior. How can we push people towards a more healthy and active lifestyle?
I focus on Quality and Wellness, so my main goal is to identify the social factors that prevent people from getting the healthcare they need. To be more specific, let’s say I have a machine learning model in my solution that predicts a certain person is due for a mammogram screening. I’m organizing a wellness initiative so I call this person in for a screening. What if they have mobility problems, that is they don’t have transportation? How will I close the so-called “gaps in care”, if they have housing and food insecurities issues? These challenges are obviously well known to that person but not to the organization. We call these “social determinants of health”, so the use cases I encounter daily predominantly deal with identifying and assessing the propensity of a patient to close those gaps and also give the physician the correct tools to have meaningful conversations with their patients to uncover these social risks.
Rosaria: Let’s talk about money, since you talked about cost reduction. How significant is the impact of data analytics in a business? What are some immediate impacts that you have witnessed in your experience?
Chenny: Good question! The US is undergoing a major transition in healthcare: we are moving from a Fee-for-Service to Value-based payment model. This means that every initiative is not just going to be based on how many times you visit the physician but also on the quality of the service you are delivering.
With Medicare, a federal health insurance, we have something called star rating, where you measure a health plan based on a scale from 1 to 5 with 5 being the highest. A challenge we might be presented with is turning a 3.5 stars programme into a 4.5 one. To achieve this, it is crucial to identify the underlying causes for the initial rating and address them effectively.
Data plays a pivotal role in this process and that’s where KNIME comes into play for us. KNIME enables us to analyze data and to perform a backward analysis, uncovering the ‘gaps in care’ that led to the lower rating, and taking actionable steps to rectify them. In this way, we are able to better the overall quality of care and make it more cost effective.
Alexandra: You have mentioned the emphasis you’re placing on building products on no-code/low-code platforms. What made you steer in this direction?
Chenny: Today every manager in every organization is data driven. They are probably proficient in Excel but new data comes in every day and Excel does not scale well. One solution would be to learn a programming language like Python or R, but because managers are busy with running the actual business, they most likely do not have time for it. They know what problems they would like to solve with automation, but they lack the knowledge on how to do it.
A typical example in many organizations is attrition, that is the gradual loss of workforce. My operation managers have a lot of data at their disposal and they have to identify the trigger factor behind attrition. By using KNIME, they are able to finally integrate the disparate data without having to spend time on learning a programming language –this is the true power of a no-code/low-code platform. In this way, they are able to educate themselves on data science concepts without the barrier of a coding language.
Alexandra: Let’s touch on the concept of citizen data scientist. What does that mean to you and your team today and how do you expect that concept to evolve in the future?
Chenny: There are so many low-hanging fruits in today’s data science world. Let’s be honest, we won’t be able to unlock the full potential of generative AI unless we are able to provide clean data. Clean data is a fundamental building block to everything and if the industry that creates data from the business processes can make sure that data is already clean, the possibilities are endless.
That’s why in the healthcare industry, we are really focusing on data governance, data management, metadata discovery and data interoperability. Once the data is clean and interoperable, then we can really start trusting the solutions offered by machine learning. Otherwise these solutions are going to make predictions based on incomplete and inaccurate data that is not really going to help anyone, resulting in a waste of both time and money. That’s where the transformation is going to happen: interoperability and data governance.
Alexandra: We have learned quite a bit about the healthcare industry and the real social impact that your product can have in addressing the “gaps” in patient care. I am particularly interested in how you’re using retrospective analysis to map prospective outcomes so policy makers can take effective action. All of this takes a lot of effort, work, and, of course, a lot of data. How has KNIME facilitated this undertaking?
Channy: Let me take a step back to two years ago and to what happened with Covid-19. In the US, we lost more than 1+Million lives. If we look at the data we see that the people who were most affected by the virus were the people on the frontline, social workers, and people who needed to take public transportation to get to work. Why? Because the education on how to handle yourself in those situations was not effective. An adult that already had some chronic conditions was not reached in time by the appropriate live-saving guidelines, for using a mask and getting vaccinated. The data was there but that was not enough. You need to turn data into information, information into knowledge, knowledge into policy and education, and finally educate the public. This loop is called “Learning Health System” and today it can take anywhere from 6 months up to 17 years. If we can reduce this loop, more and more lives will be saved.
To make this learning loop as fast as possible a tool like KNIME is crucial, not only because the information can be transmitted in real life but also because you can easily scale this from a patient to a population level. Once you are dealing with big data, you can go ahead and perform pattern and cohort identification without writing a single line of code.
Rosaria: You said you use KNIME very often, but how exactly did KNIME help you and your team in your work?
Chenny: I got exposed to KNIME at a postgraduate program with UT Austin. I didn’t expect to learn KNIME by going to this program, which was mostly centered around AI. However, when I look back at the way the program was being taught to us, I can relate to some of the challenges we are facing within our data-fragmented healthcare today. Everyday we have to work with claims and clinical data but also with a lot of third party data or open source data (e.g., census data) that is sitting out there. With so many different sources, KNIME can be used to access and blend the data in one uniform environment to develop machine learning models.
The other major application for us is data exploration and visualization. One of my favorite KNIME nodes for that is the Data Explorer node. It is able to perform univariate analysis in a matter of seconds, while with other tools like Excel it would take much longer. With KNIME you are able to import files from different data sources and understand what you are dealing with quickly and effectively. On top of this, you have everything in one place, meaning you do not need any separate tool for visualization.
Overall, I see KNIME as a Swiss Army knife: a tool that adapts depending on your needs. In my product, we are undertaking a tremendous data integration initiative, we need to visualize large amounts of data and we also need to perform machine learning on it. KNIME can help us in all of these tasks.
Rosaria: I usually ask what is the feature of the KNIME product that you use the most at work but it seems that what you enjoy the most is the completeness of the tool, am I right?
Chenny: Correct! You don’t have to use multiple tools if you have everything in one place. You have access to the data and then you just ask yourself what problem you are trying to solve. Personally, I find the node repository extremely useful, especially because it is always being updated with new nodes.
I’ll give you an extra element about KNIME I enjoy. When you are teaching, it is very easy to assume that your students already know something, but many times that’s not the case. Using a no-code/low-code tool can help closing this gap, making learning much more intuitive. I trained a group of around 30 people in my organization on machine learning and, although at first they were not able to grasp everything, once I showed them this no-code/low-code tool everything was clearer. The word spread and even some people from other departments, such as Finance or Operations started using it, upskilling themselves in data science.
Rosaria: We have interviewed a number of teachers on My Data Guest and they all said that with KNIME you have the time to focus on theoretical concepts because you do not need to invest time in learning a programming language. If I asked you your top three KNIME nodes, what would you say?
Chenny: Well, I have already mentioned it but:
- The Data Explorer node makes it to the shortlist for sure.
- All the file reading nodes, like the CSV Reader or Excel Reader nodes.
- Any visualization node, because they allow you to perform and visualize univariate and multivariate analysis in seconds.
I must add, though, that for me it is not really about my favorite nodes, but it is more about how easy it is to find the right node at the right time. Having the possibility to use keywords to look up nodes in the repository really is a game changer. On top of that, you have extensions that you can download separately depending on what you want to do. This modular approach allows the tool to be light and fast, since it is not taking up all your memory with nodes you are never going to use.
Rosaria: Since you mentioned data visualization, what is in your opinion the set of visualization plots and charts that everybody should know?
Chenny: When we are performing univariate analysis, we use the box plot and the histogram a lot to try to better understand what we are dealing with. Once we have had a look at all our individual variables, we can look at the interplay between them. Since we are interested in closing the gaps in healthcare, we are interested in variables like age, gender, race or ethnicity. We use bivariate scatter plots, conditional box plots, and parallel coordinate plots. The human brain can handle two variables but we need to be able to perform multivariate analysis in this intricate and diverse world so KNIME really helps us with that.
Rosaria: We are reaching the end of our interview. Before we say goodbye, we have one last question. Where do you see data science going after the AI big hype?
Chenny: The next hype will be about taking all these great tools, technologies and intelligences and applying them in enterprise applications. We just started exploring and working with generative AI because healthcare is a heavily regulated sector. Yet, despite the regulations, experts everywhere are trying to understand how to leverage all of that, for example for faster and better machine learning applications. Once the data is understood and its power unlocked more lives will be saved.
Rosaria: How can people connect with you?
Chenny: They can check out the Cognizant website and they can also get in touch with me on LinkedIn to collaborate.
Rosaria: Thank you, Chenny, for the great conversation and for making us aware of how data analytics and KNIME can be used in the health insurance business.
Watch the original interview with Chenny Solaiyappan on YouTube.
