The Artificial Intelligence & Data Visualization partnership
“Alexa tell me how to treat all my patients.”
That statement is naïve at best and dangerous at worst. Arguably it is also where the industry is heading: Data led informed decisions. But how do you get there responsibly?
Let’s look at today’s Pharma buzz words: Big Data, Data visualization, Artificial Intelligence (AI), and Machine Learning. We know there is a lot there to harness and leverage within Pharma – but often the biggest challenge is where do you even start?
By baking a data cake…
Layer One: The base of Data
Let’s start at the bottom – the data itself. Data is the key and is growing exponentially. This in itself creates a double-edged sword: better “potential needle” of insight, but in a much larger haystack.
Take GSK for example, they recently performed a real-world lung study in the north of England with 2,800 COPD patients. 130 pharmacies were involved across 80 GP practices, needing 3,000 people to be trained as part of the study. This resulted in over 235 million rows of data to extract insight from.
Now we have our base.
This is a pioneering approach to clinical trials, creating a large volume of non-standard data. This data closely reflects those treated in everyday clinical practice to demonstrate efficacy and safety of medicines.
So, we have the data. But now we need to understand it…
Layer Two: AI & Machine Learning
Enter Artificial Intelligence and Machine Learning. They are often used interchangeably but are technically different. AI is the broader concept of machines being able to carry out tasks in a way we consider “smart”. They replicate human intelligence. Easy example – I was researching data visualization on Google. I only got to “data v” and Google’s predictive text algorithm provided me suggestions:
Netflix uses AI to predict what you want to watch. Amazon uses AI to predict what you want to buy. Machine learning is one way to achieve AI. Arguably the most elegant, whereby we give data to machines and let them learn for themselves. This, as opposed to human’s having to anticipate every option or result chosen, building millions of lines of code with complex rules and decision trees.
This combination of AI/Machine learning is data analysis. It saves time and increases data accuracy. It shifts the legwork onto algorithms to learn and automate data recognition and identify patterns. Which in turn can lead into the world of predictive analysis.
Now we’re starting to see value. Now we’re starting to see there is a story that can be told. However, that still doesn’t mean your story will be understood by the Healthcare Practitioner (HCP) or patient, or payer. Enter layer three…
Layer Three: Data Visualization
The story behind the data is inherently relevant – if you can tell it.
Pharma has a wealth of data from multiple therapy areas and multiple studies with hundreds of patients. Typically, MSLs would only be able to present the summary level of efficacy etc in these clinical trials to HCPs. The raw data behind it was too large to easily drill into and understand patterns. Applying AI gives us the ability to group and sort the data into logical data sets – e.g. understanding the impact of the studies by age, by gender, by type of patient (naïve vs switch).
Introducing data visualization gives a way for the MSLs to intuitively present that data to their audience, so they can investigate and understand those patters themselves. This resulted in one of the most impactful responses from HCPs:
It gives me deeper answers to questions I want to ask.
Data visualization allows a story to become fluid, personalized and exploratory. What was once a linear series of slides in an eDetail now becomes an interface where a HCP can choose to deep dive into against a set criterion of filters they choose in real time.
This results in a greater level of transparency and trust between Pharma companies and physicians. There is no way to hide the data – the results are the results.
So, we have the data. We have analyzed it. And presented it. Is there anything else we can do with it in the future?
Layer Four: Voice Search
People have no time. Physicians are no different. They want answers to their questions now.
“50% of all searches will be voice searches by 2020” according to comScore.
Going back to our first statement: “Alexa, show me how to treat my patients”. In the context of this “Data Cake” we can now see how the pieces fit together. If you can have a data set which has been confidently cleansed and approved, and if that data set has been accurately grouped, and if that accurate grouping and analysis has been visualized in a way to answer a multitude of questions a HCP may have then the next logical step is to find the easiest way to present that data when requested.
By understanding the questions HCPs have and building data visualization responses, using another layer of AI or machine learning, we now can easily and confidently present accurate, and even predictive analysis and treatment messages to HCPs; giving insight that historically may very well have been lost.
This of course must be balanced against clinical expertise and will never be a substitute for traditional doctor/patient interactions, but strongly supports the physicians need of informed decisions when prescribing.
Here at PharmiWeb Solutions, our innovation team are always looking to future. Get in touch to bake your own data cake.
Doctor Alexa will see you now.
ABOUT THE AUTHOR
Dave has driven award-winning digital campaigns across a range of Blue Chip Pharma, Technology and Professional Service clients. Dave has over 10 years' experience in digital marketing.