AI Expert Perspective Interview with Kyle Flickinger

Recently Kyle Flickinger, our Vice President of Bio Pharmaceutical Solutions, was interviewed by AI in Pharma for their upcoming summit. In this interview, Kyle discusses the advancements of AI in Pharma. 

Kyle will be sharing more of his thoughts in Boston this October 9th at the AI in Pharma Summit 2019. He is joining the panel discussion at 12:20PM: “When and Where is the Right Time to use AI?”

You can read the interview below:

Kyle Flickinger


What conversations need to happen to help advance the field of AI in Pharma?

In oncology, we need to close the gap between where and how often precision diagnosis is happening. This is a significant imperative in the community setting, where timely patient treatment decisions are being made. These are needed conversations between providers, community affiliates and study teams implementing therapies that challenge the diagnosis and treatment paradigm. The conversation about mapping the “patient journey” needs to evolve and become a dynamic part of how AI is configured to support more intuitive workflows at the provider level. There are fundamental reasons why only 5% of cancer patients participate in research. Patients and care teams are not being identified and presented with research-based care options in the small windows of opportunity where these decisions are made.

Where do you see the industry in 5 years time?

My hope is that AI is an integral part of timely precision diagnosis and that we see more rapid adoption of molecular testing by community oncologists. As a result, we should have more first line precision oncology therapies being made available. We need platforms that can easily embed AI-based translational models into workflows across a wide ecosystem of providers and affiliates. How we get there is another story. By some estimates, the AI healthcare market is projected to grow to almost $40 billion by 2025. Additionally, some analysts believe that hospitals and physician providers will be the major investors in machine learning and artificial intelligence solutions and services. We are already seeing this over reliance on hospitals and providers to fund these efforts being challenged.

What is the biggest limiting factor holding back AI? How are you working to overcome this?

There are many limiting factors holding AI back at the point of care. One of the biggest we see is seamless implementation of new algorithms into the care workflow without disrupting standard of care. Oncologist and care team burnout continues to rise as well as frustration in keeping up with new advances in precision medicine. Deep Lens has a unique perspective on these obstacles holding back AI in precision diagnosis. Our customers are both the provider networks including pathologists, oncologists, clinical care teams and bio-pharmaceutical clinical development and brand teams. In many ways, creating visibility into ongoing challenges from both perspectives is helping to bring awareness to the community setting and molecular testing with AI. Collectively, the interactions and input received from these groups have been integrated into the development and implementation of our VIPER platform. We are continuing to create and refine solutions where new AI models from translational teams are embedded to transform how care teams collaborate; empowering quicker, more precise treatment across all cancer types.

What is Deep Lens doing differently?

As referenced previously, we do not believe AI in healthcare will scale if we expect providers to be the investors of time and capital. Deep Lens has made a commitment to improving provider-patient care coordination through the broad use of our free VIPER platform at institutions across the globe. However, just being free is not enough to move to the top of the strategic list of imperatives. We are working to solve many fundamental issues in precision diagnosis workflow to drive adoption of AI. These aren’t always the most flashy or obvious issues, but being able to process 10% more cases a day for a pathologist, or helping a research coordinator communicate to an oncologist a required molecular test/result at time that ensures a patient meets the final inclusion criteria to enter a study are critical.


What is Deep Lens working towards in the AI space?

Deep Lens has a number of initiatives utilizing machine vision based algorithms from different industries that in some instances require less data for ground truth. These are specifically focused on helping translational and pathology groups to improve day to day workflow as well as more advanced projects to determine prognostic indicators of outcome as early as possible. One of the aspects of our model that is very attractive to these groups is our ability to deploy AI in a clinical/commercial setting while maintaining exclusivity and ownership to the creators of the algorithms.


When do you believe we will see the true capabilities of artificial intelligence?

For Deep Lens, we are mission focused on improving the participation rate of patients in precision oncology studies. A secondary value of our approach is developing AI based methods early in the development cycle to identify patients who should be offered molecular testing. In many cases, the care pathway requires sub-optimal first line treatments and progression to determine when a specific diagnostic should be run. For us, we will see the value and true capabilities of AI when we are able to accelerate and optimize this process driving more precision based first line treatments as standard of care.

What are you most looking forward to at the AI in Pharma Summit?

I have participated in a Grey Green Media event in the past and it was a highly focused and interactive day spanning many different stakeholders across the healthcare ecosystem. We need more of these interactions if we are to solve for some of the obstacles preventing AI from being effective at the point of care. Based on the agenda and speakers, I am encouraged that the conversation will span AI in translational science through to how these models can be rapidly implemented to support patient identification, diagnosis, and care planning workflows.

This article was originally published via

Learn More About Kyle Flickinger

Download Our Brochure