Expert Spotlight: Renu Gehring – AI “A Set Of Russian Dolls”

Renu Gehring, a technology leader with 20 years of experience in creating healthcare solutions that solve healthcare problems, talks about data science and technology in healthcare, leading from the front and 5 key skills for succeeding as a healthcare CTO. Connect with Renu on LinkedIn here.

1. How do you see the role of technology evolving in patient care over the next decade?

I want to begin by redefining technology to encompass data science tools and algorithms, including Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL). DL has advanced machine understanding of human language, as demonstrated by large language models. It has also driven AI-enabled drug discovery, exemplified by DeepMind’s AlphaFold.  

So now your question is: How do I see the role of data science and technology evolving in patient care over the next decade?

Short answer: I believe that data science and technology in health tech will evolve towards greater precision and specificity. The era of one size fits all is gone.  

Let me elaborate by highlighting how data science and technology will evolve in patient care at two contrasting ends of the spectrum: primary/preventive care and complex care.  

1. Primary/preventive care is characterized by volume; everybody needs it.  The pandemic demonstrated that primary care is effective in virtual settings, creating a massive shift of healthcare into our homes and driving innovation among health tech companies.  Soon it is likely that we will be taking our vital signs with our phones!  Many healthcare providers are already investing in and implementing Ambient AI, which works in the background, to enable efficiencies for physicians, nurses, and staff.  What does this mean for technology and data science? It will mean choosing the most appropriate technology and data science to fit specific use cases.  A healthcare company will probably want to pursue both cloud and on-premises storage and compute and invest in small language models addressing different use cases.  Surely the AI model enabling provider-patient communication will be different than the one automatically verifying patient insurance and benefits information.
2. Complex care especially in oncology and rare diseases. Data science and technology are spurring innovation in complex care, leading to increasingly precise and specific models. AI enabled drug discovery is enabling highly customized, targeted treatments. I believe and hope that the trend towards precision medicine will continue to accelerate in the next decade.  I also believe that AI/ML/DL in drug discovery, ML/DL in real world data (including biomarker and genomic data) will lead the drive towards specificity and precision.

2. Which emerging technologies do you believe will have the greatest impact on healthcare in the near future?

I have already talked about the importance of data science tools, including AI, ML, and DL. I certainly believe that these and their supporting technical platforms like Databricks and Snowflake will have a outsized impact on healthcare. I believe and hope that they will help healthcare providers and administrators do their job better, faster, and more efficiently, leading to better patient outcomes and satisfaction.

I believe that there is a gap in the availability of tools to standardize healthcare data for analytical purposes. Over the past 25+ years, most of my work has focused on extracting and transforming data to make it analytics ready. This is because healthcare data is complex, vast, and disparate and not designed for the purpose of analytics.

Take healthcare claims or electronic health record (EHR) data as examples. While both claims and EHR are standardized datasets, their primary purpose is payment and delivery of care respectively. Substantial effort is dedicated to preparing these datasets for analysis. This is a paradigm I have witnessed repeatedly in my career.

I would love to see tools for reducing the level of effort in getting healthcare data ready for analytics.  To that end, I am closely following the common data model and healthcare vocabularies created by Observational Health Data Sciences and Informatics (OHDSI) initiative.

What role does/will artificial intelligence play in your current and future projects?

First, let me clarify that I view Artificial Intelligence (AI) as a broad field that includes Machine Learning (ML) and Deep Learning (DL).  Think of it like a set of Russian dolls. AI is the largest doll, containing ML as the next smaller doll.  Within ML, there is another small doll called DL. 

ML has played a star role in my past projects, among which two stand out. The first involved the prediction of Emergency Department utilization to move patient care to urgent or primary care modalities. The second project involved categorization of patient emails for improved patient retention and satisfaction.  

For the past 2-4 years, I have been leading initiatives where DL is a key element. One notable example is the usage of transformer models to predict quality in drug manufacturing. I am planning to embed Generative AI in many processes to drive efficiency. To create higher value for our clients, I am also leading several initiatives that include ML on patient prediction, physician targeting and next best action.  

Can you share an example of a major project or initiative in healthcare technology that
inspired you?

Drug discovery is an inspiring and extraordinarily complex field. It generally takes collaborative teams of chemists, biologists, bio-informaticians, data engineers, and data scientists an average of 10-12 years to progress from drug ideation to FDA approval. The process begins with identifying a target, sometimes a protein, in the human body that is involved in the disease pathway.  

Recently I concluded an initiative to mine new patents enabling chemists to select promising targets for drug development. The initiative was challenging from both IT and domain knowledge perspectives. To contextualize, around three million patents were filed across the world in 2021. 

Our task was to ingest and analyze incoming new patents while also distilling pertinent information from four decades of patent history. Our team of chemists closely collaborated with technology and data science teams, sharing insights into chemical formulae interpretation, disease etymologies and pathways, and drug mechanisms of action.

Despite working with lean teams spread out across three continents, we achieved our objectives through the combination of machine learning algorithms and human led supervision. These algorithms targeted patents of interest and enabled us to find the proverbial needles in the haystack.

How do you foster a culture of innovation within your technology teams?

My past and current experience has been leading teams of data scientists, data engineers, ML-Operations engineers, platform administrators, and software engineers so I will talk about fostering a culture of innovation in this context.  

1. Encourage curiosity within any role. Ask open-ended questions that gets people to realize that they have agency in solving their problems.  
2. Lead from the front. This has served me well because I love to learn and write code. There is such beauty in Python! I feel that people respect a technical leader who is eager to walk in their shoes and learn from them. Besides, coding is such fun with AI enabled coding assistants.  
3. Encourage a plurality of opinions and invite criticism. Both help to create trust and a safe space for innovation.  

What are the key skills and qualities you believe are essential for a successful CTO in the
healthcare tech sector?

1. Have high emotional intelligence. Lead with heart and good intentions. This means being kind, expressing empathy, and genuine respect.  
2. Be a visionary. If you want to inspire people, you need to have a vision that you co-develop and implement with your teams.  
3. Have a growth mindset.  Constantly learn and teach others.  
4. Be humble yet confident.  I know oxymorons.
5. Technical expertise. This is a must have because no one tool or model or platform can fit all your company’s needs. You will need to be discerning enough to build/buy solutions that are appropriate for your company. 

Renu Gehring is an AI/ML and technology leader with a proven track record for creating solutions that solve complex business problems.  With over 20 years of experience in healthcare, Renu is a published author of two books on healthcare analytics with SAS Press.  Her current role is leading technology and AI/ML enabled initiatives to support growth at CareSet.  Previous solutions that she has led include AI/ML algorithms to improve drug quality, ML enabled mining of patent data to support drug discovery,  and several Gen AI initiatives to increase productivity and efficiency.  Renu loves to connect over all things technology and data science and can be reached through her LinkedIn profile.