Ylan is a Data science executive with 8 years experience leading high performance analytic and machine learning teams. He is currently the Vice President of Data Science and Machine Learning for UnitedHealth Group and leads a team of data scientists that focus on improving health outcomes for patients.

Justin: What AI/ML/Data based solutions are currently successful in healthcare? 

Ylan: The first solution I’ve seen that has been successful is using AI on medical imaging, particularly to make diagnoses. Deep learning has led to significant advancements in predicting a host of different diseases and in some cases outperforming radiologists. I would expect to see continued advances in deep learning image recognition. 

Another area I’ve seen success is with risk scoring patients for specific disease states. In the past, patients would get a certain disease and then the disease was treated. With different machine learning algorithms, it is possible to predict whether a patient will get a specific disease and to determine how successful clinical treatments will be in managing the disease.

Justin: What do you see as the main blockers to really progressing these and other solutions being implemented? 

Ylan: The two main blockers that I see are people and access to the right data. From a people side, there are varying levels of knowledge for what AI can be used for and how to effectively implement it. Without C-Level support, AI cannot be successful in any organization. To get that support, we need more leaders that understand AI from a business standpoint and how it can be an important tool to use in the right situations. 

From a data side, there is more data out there than anyone can analyse. We need to be better as an industry in identifying valuable data assets and make these assets available at an enterprise level. Too many times, the right data exists in an organisation, but few people are aware of it. This acts as a barrier to leveraging the full power of AI.

Justin: There are concerns in the general public around AI ‘ethics’ and data privacy. What steps could be taken to make the greatest difference to the public’s perception? 

Ylan: This is big challenge, as AI ethics are still in their infancy and haven’t been clearly defined. What I find interesting is that people have given up their data and privacy in exchange for using “free” services like social media. As consumers have become savvier, though, they’ve started to realize that they may have made a mistake. There is a privacy and ethics movement that has started to gain strength, and we’ve seen companies like Apple attempt to be more transparent in how data is used. 

Changing the public perception of AI ethics and data privacy will require a much greater degree of overall transparency. I also think that individuals should have ownership over their data and be allowed to sell it just like any other market. Our individual data is valuable and shouldn’t be given away for free. We also need to define an appropriate system of AI ethics, specifically what the limits are and if anything is completely off the table. I think a detailed system would change the public perception of AI and give people confidence that it is being used ethically.

Justin: What do you see AI/ML/Data doing for the healthcare industry in 5 years and that we as the general public will take for granted? 

Ylan: Improving the quality of healthcare delivery and helping to reduce costs. The healthcare experience hasn’t changed much for the patient over the last ten years. We still generally will go to a clinic, spend some time in the waiting room and then get to see a doctor for a limited amount of time. Using AI in partnership with clinicians will make this entire process more efficient and allow the clinician to spend the majority of their time delivering care to the patient. Things like documentation, billing, and scheduling will be majorly streamlined using AI. 

There is a lot of waste in the system that is very costly. As machine learning is applied more to fraud detection and even the improvement of clinical best practices, this will lower the overall cost of care delivered. I would also expect to see more standardization of care, so ideally it won’t matter if you receive care in Chicago or in a small town in Wisconsin. The quality of care should be standard whether you are in a rural area or urban metro, and AI can help to make this a reality. 

The public will eventually take this for granted. If you compare the user experience of someone using Amazon or Google services, it doesn’t compare to the current state of healthcare. There is a lot of catching up to do within healthcare, and as the consumer starts to get a better experience they will also start to demand more. Many of the AI innovations that will be implemented will be behind the scenes and the patient will be unaware of them.