Last Updated on August 9, 2024
Alpa Reshamwala’s career has been a fascinating one, and it’s also a career that has tracked along with the widespread adoption, and subsequent development, of data mining techniques and technologies.
Going back to the start, Reshamwala was working as a lecturer at the Babasaheb Gawde Institute of Technology when she started teaching classes on relational databases. This was the start of her enduring interest in data-centric work.
In 2006, Reshamwala enrolled in a data mining class while earning a graduate degree at Mumbai University, Thadomai Shanai Engineering College.
In 2009, she was offered an Assistant Professor position at NMIMS University, Mukesh Patel School of Technology Management and Engineering. This is where Reshamwala delved much more deeply into data mining, publishing research papers and testing her data mining skills by examining student test data to determine likely at-risk students to improve student retention.
Reshamwala also applied her data mining skills to computer network data at the institution to bolster security against malicious DOS (Denial of Service) attacks.
Then, in 2018, Reshamwala started working with non-clinical patient data sets with a number of organizations, including Aristotle Psychological & Biofeedback services, Neighborhood Psychiatry Associates of Manhattan, and CVS Pharmacy.
This work with healthcare-related data inspired Reshamwala to explore data mining possibilities in healthcare, which uncovered a significant number of opportunities for application and improvement.
At its core, data mining is about searching for patterns within large datasets, and the healthcare industry naturally has enormous amounts of patient data. As Reshamwala explained during our visit, that same patient data, when mined, can be used to predict future outcomes.
This is exciting news for patient care. With rigorous data mining, patient diagnosis could be substantially improved leading to better care regimens. If healthcare professionals know more about the experiences of other patients with the same condition and how those patients responded to different forms of treatment, they can make highly informed decisions about the treatment of their own patients.
But healthcare data mining doesn’t stop there. Insurance applications are also viable, with data mining helping to find patients the best coverage for their situation. Stored patterns can also expedite the claim adjudication process.
There’s more:
“Data mining has applications in pharmacy, lab reports, patient accessibility, and providing the best customer service.”
Reshamwala took the time to guide us through healthcare data mining work and the effects of applying this technology. Let’s begin.
Healthcare data mining vs. mining in other industries
One of the questions we were most eager to ask Remshamwala is whether there are major differences between data mining in the healthcare industry compared to data mining in other major industries.
We’ve already noted that healthcare institutions amass large amounts of data. By law, records for Individual patients must be kept for at least six years, with some states requiring a longer period of retention.
What Reshamwala pointed out to us is that the data itself is quite different compared to data that other industries are looking to mine.
“Data mining in healthcare is more critical since it involves patient care information and personal information. Patient privacy and confidentiality are very important, and data mining involves various strategies to comply with this requirement.”
Any mishandling of personal data and medical data can lead to legal action, making it even more important for healthcare data mining to stay highly secure.
Reshamwala also noted that medical data is much more complex than data from other industries. There are many different dimensions to this data and potential outcomes, even in instances where patients share multiple data points.
So to summarize, in healthcare data mining, the data is hard to wrangle, and it’s incredibly important that the data is handled carefully.
Current work
When looking at the specifics of Reshamwala’s day-to-day work, it’s important to acknowledge that mined data isn’t terribly valuable on its own. Gathering this data is just the first step of the process.
Data can then be staged as descriptive analysis, predictive analysis, or prescriptive analysis, and each of these can lead to notable discoveries.
Still, Reshamwala’s main focus is on the mining process itself. Speaking about the specifics of her current research, Reshamwala had this to say.
I am currently exploring and concentrating on optimizing revenue cycle management by mining sequence patterns along with efficient diagnosis decisions, leading to high-quality patient care.
Reshamwala has also been able to leverage the lessons learned from previous work and research experience. As an example, during her time with Aristotle Psychological & Biofeedback services, she explored how sequences of The International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes could be auto-filled based on a combination of the primary ICD-10 code, the Current Procedural Terminology code, and the specific insurance being billed.
The result is reduced billing errors, which means better accuracy and timely reimbursement from insurance providers.
This concept was further developed through Reshamwala’s work with doctors at Columbia and Mount Sinai, adding more complexity to data analysis and the strategies for prediction.
Reshamwala also worked on analysis-based prediction for CVS pharmacy, this time looking at stock registry and customer prescription refills for chronic illnesses.
Reshamwala has made it very clear through years of dedicated work that she’s looking to improve numerous aspects of healthcare, not just one small component. And while approaches for one of these areas can’t be transferred to another area wholesale, there are definitely lessons learned from each experience and plenty of opportunities to apply those lessons to other applications.
A notable advancement
As for Reshamwala’s professional contributions to data mining, one of her most notable innovations is her incorporation of time intervals when looking at patterns.
“Artificial intelligence techniques were applied for higher accuracy in predicting future occurrences of the patterns. After learning about the current state of the data with previously stored time interval patterns, models can be developed for predicting future patterns with the time interval on its occurrence.”
Adding a time component to data mining, rather than only looking at patterns, adds value to the analysis of that data, and, as Reshamwala touched on, it can also enable prediction models, the value of which should be self-explanatory.
Making this even more exciting, the potential applications of time interval data mining and analysis in applications beyond healthcare are substantial, and we have Reshamwala to thank for pioneering this approach in her research and professional experience.
Applying prediction models to a customer service context, for example, could mean that CX teams would be able to reliably predict support and maintenance needs.
For retail sites and physical stores, prediction models could forecast demand for certain product categories and/or foot traffic based on the time of day, the time of the week, or even the time of year.
To be clear, just because it’s now possible to add a temporal element to data mining doesn’t mean that it’s easy to achieve. This is difficult, highly technical work, and Reshamwala’s breakthrough in this area was the result of many years of study and real-world experience.
As our visit with Reshamwala came to a close, we couldn’t help but wonder whether she’s planning to continue working in healthcare or take her talents elsewhere.
Looking forward
So will Reshamwala stay dedicated to data mining in the healthcare industry? The short answer is yes. Here’s the long answer:
“Yes, I will continue to work in healthcare, with its wide variety of dimensions and open issues, as well as exploring huge data with combinations. I’m currently also exploring data on oncology patients and the range of applicable strategies. There’s also ongoing parallel research regarding pandemic patient care optimization due to COVID-19, which can also be applied to all variants in the future.
Our single biggest takeaway from our visit with data mining expert Alpa Reshamwala is that healthcare is inherently complex. The data is complex, processes are often complex, and networks of institutions, care teams, insurance providers, and billing professionals all have to work together and communicate efficiently with patients to determine any current or potential future health issues and seek out the best care options available.
In reality, the healthcare industry can seem less like an industry in the traditional sense and more like a delicate ecosystem.
Further, from the perspective of the average US patient, the US healthcare system is certainly in need of improvements and optimizations. While data mining may not offer solutions for every systemic issue, it’s rapidly proving its worth on many different fronts.
Skilled data mining pros are putting this tech to work and finding new ways to leverage impressive capabilities, and as long as they proceed with care, millions of patients stand to benefit from their efforts.
