Meet the Expert: Arin Brahma, PhD

Assistant Professor, Loyola Marymount University

Arin Brahma is both a generalist and a specialist with diverse experience that spans the depth and breadth of IT and AI applications as well as a seasoned educator and consultant. He is currently Assistant Professor of Information Systems and Business Analytics at Loyola Marymount University (LMU) in Los Angeles. Arin teaches AI, Big Data, and Operations Management to undergraduate, MBA, EMBA and MSBA students. He also serves as the founder and CEO of Kognivo, an AI consulting company that solves industry problems by building functional solution models using research-grade AI/analytics, design skills, and client data. In our interview with Arin, we asked him about the future directions of AI in healthcare and learned about the use of this discipline in an innovative student project he is directing at a major Los Angeles hospital.

AI is a hot topic in healthcare right now. Where do you see it going today versus the future?

AI has advanced in the areas of disease diagnosis such as scanning x-rays to detect lung cancer. Use of AI for diagnostics is highly impactful. However, AI also has great potential in improving operational efficiencies such as determining accurate staffing levels to reduce costs. Imagine a hospital ward with 100 patients with different diseases at various levels and you don’t know what the patient composition will be with respect to the diseases states seven days later. If that can be predicted, then appropriate resources and specialized staffing can be scheduled. It probably won’t be 100 percent accurate but even if it is 80 percent accurate, that’s a huge deal.

AI, working in conjunction with cloud computing, can also enhance interoperability – a central focus of the Affordable Healthcare Act. If all healthcare systems can communicate with the cloud, then they can communicate with each other. While security breaches are a major concern, hospitals understand the importance of taking advantage of cloud computing and how much money can be saved through improved efficiencies.

What do you think it will take for health systems in mass to buy into the cloud?

Historically there has always been a natural resistance to changes that automation and technology bring, and the healthcare industry traditionally has not been an early adopter of new technologies. Probably the biggest barrier for healthcare CEOs are security and privacy concerns. However, as they see other industries taking advantage of AWS or Google Cloud and see success, I think that will go a long way in healthcare embracing these technologies. It’s already happening and it’s just a matter of time for them to take advantage of the AI, big data, and cloud technologies in a pervasive scale.

What industries are doing well at adopting these technologies? Who should healthcare CEOs be watching?

Financial services, retail, travel—they’re doing well. While hacking concerns are still an issue, in today’s world Zelle money transfers have become routine. There’s no concern about the money going to the right person or not. Financial services have been a leader in building that confidence. Same with online trading. Not that they haven’t been hacked, but they recover and get stronger. Healthcare should be looking at this and taking its cue.

Describe the Loyola Marymount University MSBA (Master of Science in Business Analytics) project that students are working on for a Los Angeles hospital.

In hospitals, clinical volume goes up and down, sometimes very unpredictably. With COVID, virtual processes and telemedicine have added a new variable. The project’s goal is to apply modern AI techniques to predict, as accurately as possible, what will be the weekly clinical volume which in turn impacts staffing levels and operational costs. Without more definitive visibility into upcoming volume, unnecessary costs occur due to under or over projection of hours and staff. In the case of understaffing, patient care quality suffers and overstaffing leads to increased healthcare costs. With such advanced technologies, accurate projections can be in real time and staff scheduling adjustments can also be automated based on what the future volume will be. A successful implementation of this project will have tremendous impact on the hospital’s patient care quality and the bottom line.

What parameters do you look at to predict what will or won’t happen?

In the beginning, we don’t know because as humans we are not really good at understanding data patterns. Computers and AI algorithms are very good at that. So, the idea is to put as many features into the model as possible – even all possible demographics, social factors, and geographic features such as zip code. AI is capable of analyzing the data to see which factors emerge as having the most influence. It can also find out how those factors can be combined to create as accurate prediction as possible. In the beginning, we don’t know which features will be the strong predictors. Sometimes we’re surprised. It’s a discovery process. Ultimately, the goal is to translate the AI model findings into good operational controls that will lead to a successful on-the-ground predictive clinical staffing capabilities.

What does the MSBA course work look like for the students at LMU?

The actual project is completed during the summer semesters with two sessions of two months each. However, preparation for the project starts in the Spring semester. During this time, the project is defined in detail in discussion with the client; all formal paperwork and data security mechanisms are put in place; initial ideas about the solution are brainstormed; deliverables are defined and a detailed project plan is prepared. By mid-semester, students must have a full understanding of the project and make a presentation to the client. They must be able to describe the approach, obtain feedback from the client to make refinements, and be ready to begin the project in May. Solution results are available by mid-August and a final project presentation is made.

What happens after the students recommend the solution?

Usually, all models, code, and documentation are handed over to the client along with the recommended solution. Metrices are also put in place to show the before and after impact of the solution. After this point the client’s IT department takes it over for real-life implementation.

Is there anything else you would like to add?

Yes. Recognizing that professors have extensive knowledge that can be beneficial to industry, I started a consulting company – Kognivo, for the purpose of promoting knowledgeable educators to industry, especially in the fields like AI, big data, and cloud computing. Professors are involved in cutting-edge research and know how to solve problems and traditionally report their solutions in journals. I think industry doesn’t take advantage of all of the knowledge and solutions locked in the academic journals and is missing out on the resources of academia. With an understanding of both the PhD language as well as the IT/industry language from my 20+ years of experience in the IT/industry, I’m able to bridge the gap by linking appropriate PhD’s to industry problems that need to be solved. These are our projects. We love building models, solving real-life problems, and going on to the next one. We don’t do IT implementations. If the client doesn’t have an IT department, it can be outsourced to our partner, Speridian Technologies, based in Irvine. With this concept, my colleagues and I have the freedom to solve problems without the typical company structure, since PhDs are working as a team of consultants collaborating together to solve problems, and not as employees of an IT company.