Ylan Kazi, Vice President of Data Science and Machine Learning at UnitedHealth Group (UHG) joined the Data Science for All Community last Thursday to share his thoughts on AI Trends and the Future of AI in Healthcare.
Fellows from both Data Science for All/ Empowerment and Data Science for All/ Colombia took a break from their current studies to look beyond the classroom and envision the applications of their new data skills in the Healthcare space.
Ylan is an industry leader whose team of high performing data scientists use machine learning to improve outcomes for Medicare patients. In addition to his perspective on the future of advanced data in healthcare, Ylan shared his personal career journey and how his experiences as an undergraduate Philosophy major, a masters in Public Health, and experience consulting on the business side of healthcare led to a senior data leadership role at the largest insurance provider, and second largest healthcare company in the world.
Ylan describes his path as ‘unconventional’, a descriptor we hear frequently from Data Science for All speakers. While many believe the path to a data science career requires a ‘conventional’ background of a Computer Science undergraduate degree followed by advanced degrees in a speciality field like Machine Learning or Natural Language Processing, the reality is that a diverse perspective can be a competitive advantage for a data scientist’s career. Data Scientists like Ylan use their ‘unconventional’ backgrounds to think about problems differently, and as they grow their core data literacy skills they often find opportunities that might not be suited for someone who is solely focused on neural network research, for example.
“Mentors are just fantastic because they can really give you the unfiltered truth. A good mentoring relationship is very transparent. Mentors can help you avoid many pitfalls- through their experience they’ve made many mistakes and by explaining my challenges and my situation they can help me avoid making the same mistakes in my own work.”
Ylan has been part of the Data Science for All mission since DS4A Empowerment kicked off in October 2020, serving as one of the program’s distinguished mentors. Ylan has dedicated his time coaching DS4A Empowerment Team 79 as they work to help Code Nation, a non-profit that provides computer science education for students from traditionally under-resourced schools. For its DS4A Empowerment Capstone project, Team 79 is using data science and machine learning to help Code Nation make better decisions regarding its allocation of resources.
Ylan shared that one of the key themes in his sessions with Fellows is the difference between ‘Data Science Theory’ and ‘data science practice.’ Ylan discussed, “In the real world, what these teams have encountered is you do not just go from A to Z in a linear fashion. In the course of a project, there are many times when you have to take a few steps back to reframe your hypothesis or problem statement as the business problem at hand changes.”
Ylan’s motivation to be a good data science mentor to Fellows is inspired by the data science mentors in his own life who have helped him achieve his career goals. Ylan discussed the contribution of 2 Mentors specifically, including an internal Mentor at UHG who has helped him navigate the UHG organization and bounce ideas back and forth before formally pursuing projects or techniques. Ylan’s other mentor is his source for inspiration regarding cutting edge research in AI/Machine Learning, Cognitive Computing, and Quantum Computing, pushing Ylan to apply new tactics and techniques to solve today’s business problems.
Correlation One recently released a whitepaper, The Top 100 Dream Employers: Where Diverse STEM Professionals Aspire to Work based on the data we collected from over 8,000 DS4A applicants. In terms of destination industries, Healthcare was a top choice. For this reason, we were excited to lead off our DS4A Webinar series with Ylan. However, ‘Healthcare’ is a super-industry, encompassing many different organization types and disparate roles. Before diving into Ylan’s view on the future of AI in Healthcare, we asked him to orient our viewers on the Healthcare landscape and where his work fits within the bigger picture:
“I traditionally think about my side of healthcare as the payer side of Healthcare. From a payer side, what we are looking to do is ensure that our patients are getting the most value for their money. For consumers, or in this case patients, Healthcare is different from spaces like retail where the costs of goods and services are transparent. There’s tremendous variability in the cost and quality of care from region to region, and sometimes even from neighborhood to neighborhood. Our work is a key part of a larger data science ecosystem that includes healthcare providers, pharmaceutical researchers, and electronic records/health tech companies.”
With that orientation, Ylan continued to share what he sees are the most exciting opportunities for AI in Healthcare, including his own sector, Healthcare Providers, Health Tech companies, and Research Pharmaceutical organizations:
“Within [the payer side of Healthcare], it’s using AI to reduce costs and improve quality. The beauty of AI is it that it can take giant amounts of data and it can form insights that can then be used to predict human behavior. For Healthcare Providers, there’s a really important mandate to use AI to augment doctors’ and nurses’ capabilities. There is a shortage of quality doctors, especially in rural areas. How can we help doctors and nurses see AI as a partner versus something that’s going to replace them? I don’t see robots replacing human providers, but I do see AI as a key helper to scale the impact of talented doctors and nurses.
In the Health Tech space, I think AI can do a really good job of personalizing medicine. Wearables like Apple watches and sleep rings produce data that can be used to create personalized health journeys between patients and their doctors. Today, there’s a growing demand for this.
Lastly, in the Research Pharmaceutical space, I’m really excited by the possibility of applying AI to genomic data. I think this goes hand in hand with personalized healthcare. Wouldn’t it be great if your antibiotic was formulated specifically for your genome? The exact dosage, minimal side effects- isn’t that a cool idea? With the proliferation of AI in healthcare I see this coming sooner than later.”
In 2021 and beyond, the data science world needs to understand and further the cause for Ethical AI. It is no longer sufficient to ask ‘What can AI do?’- we must ask ourselves ‘What should AI be used for?’ and perhaps more importantly, ‘What should AI not be used for?’ As a data science leader and a classically trained philosopher, we asked Ylan for his take on Ethical AI today and for the future:
“From a pitfall standpoint what AI has revealed to all of us is: if AI is using biased data, AI can amplify that bias. This concept is not really well understood. Many think, ‘If I just get enough data, when I run it through an algorithm it will make great predictions.’ But what if the data that your algorithm keeps ingesting continues to have biases embedded within it? There are a ton of examples of this.
Ethical AI is something organizations really need to think about. Companies may have access to millions of data points, but what if an aspect of that data is suspect or biased? That’s going to lead to some type of biased solution unless a lot of different techniques are used to understand and combat the underlying biases.
From a business lens, for any type of AI transformation, you have to have your C-Level executives on board. Executives need to understand not only what AI is, but how it can be used. In many cases, you hear the hype: ‘AI can make anything better’. That’s not true. It’s one of many tools and if it’s used correctly it can make things 10x or potentially 100x better.
Lastly, the ethical component. With AI, it is very easy to scale. With a company like Facebook or Google, we’ve already seen what can happen when AI has unintended consequences. For Facebook, AI can amplify conspiracy theories and hate speech because the algorithm is trying to prioritize pageviews [without consideration of the ethics of promoting certain misinformation/world views].
It’s in every company’s best interest to understand the ethics of AI and there is much more work that needs to be done. In Healthcare, if a company’s AI products turn out to be unethical it can really damage their brand.”
Before taking questions from our live audience, we asked Ylan to imagine he was the DS4A Mentor to the hundreds of viewers in attendance and share some guidance with our Fellows as they embark on their personal data science career journeys. Ylan shared 3 maxims:
Watch below to learn more about Ylan’s day-to-day role and responsibilities, the role of Citizen Data Scientists in the future of AI-driven healthcare, and Ylan’s answers to questions from the audience.
Based in Minnetonka, Minnesota, Ylan is active in the data science community and serves as an AI/ML advisor to Smart Steward, a company that provides solutions to combat COVID-19 and antibiotic resistance. Ylan also writes about how AI will affect humanity at discoveringai.com.