5 Ways machine learning can contribute to life sciences organization success

Machine learning has become one of the hot topics for discussion in healthcare over the last few years. And for good reason.

Unlike many technologies that seem to come and go on a whim, machine learning offers real promise and tangible benefits in helping all types of organizations take better advantage of their massive stores of data by gaining insights that wouldn’t have surfaced otherwise.

These insights help all healthcare stakeholder make better decisions faster, either to take advantage of unseen opportunities or to avoid costly missteps. Machine learning is also one of the key components of the larger arena of digital intelligence, which combines data and domain knowledge to create context that enables organizations to outperform the market and their competitors.

Its primary benefit is that it can sift through those massive stores of seemingly unrelated data at lightning speed and discover patterns and trends that wouldn’t have been obvious to a human. Of course, a human with domain knowledge is still required to determine whether those discoveries are worth pursuing further.

But the point is machine learning can find them, which is what is making it so attractive to life sciences organizations looking to solve business problems. Following are some of the ways machine learning can help those organizations achieve greater success.

Proving the value of a drug
In the old fee-for-service model, success in controlled clinical trials prior to FDA approval was enough for a payer to justify adding a drug to a formulary or approving a device. As healthcare continues its transition to value-based, outcomes-oriented care, however, that is no longer the case.

Today, payers increasingly want to see the value in a real-world setting. Machine learning provides those answers by combining medical and pharmacy data. It then shows how outcomes, such as total cost of care, rate of inpatient admissions, and emergency department visits differ between drug A from this manufacturer and drug B from a competitor over a two-to-three-year period.

Sales enablement teams can then take that data to both payers and providers to show how that drug is proven to both improve outcomes and reduce risk for different populations, creating a much easier sale.

Taking advantage of windows of opportunity
Two of the most important periods in the drug lifecycle are right after the drug is launched and the roughly six months between the time it comes off-patent and generics hit the market. Life sciences organizations want to maximize sales in both.

Machine learning can help uncover the optimal target markets, such as areas or neighborhoods with a high probable concentration of undiagnosed diabetics when the organization is introducing a diabetes-related drug. Life sciences organizations can then concentrate their sales and marketing efforts to providers in those areas to get it off to a fast start.

The same is true when a drug is coming off-patent. Life sciences organizations can use machine learning to boost sales (and protect the brand) most cost-effectively by understanding which providers or patients are least likely to switch to a generic based on previous patterns so they can focus their efforts on others who need more persuading.

Discovering off-label uses
Physicians are often willing to try a drug developed for one issue to determine if it can solve another. Yet unless they write a paper or otherwise publicize their success, this new off-label use will generally remain hidden from others who could benefit from it.

Machine learning can uncover those relationships by comparing HCC and NDC codes, along with biometric data, to determine if there are correlations in use and outcomes that wouldn’t normally be expected. Once they are uncovered, the life sciences organization can look into them more deeply and test them to determine if there is a new market for an existing drug.

Digging into rare conditions
The days of blockbuster drugs that affect large segments of the population are largely behind us. The next frontier is rare diseases and conditions.

The challenge there is that the usual path of feeding data about rare conditions into regression models doesn’t work, because the prevalence of those conditions is too small for those traditional techniques to pick up correlations. Machine learning doesn’t have that limitation.

It can find the tiniest correlation needles in the largest data haystacks, creating new opportunities to address these rare conditions with treatments that have proven effective. As medicine becomes more personalized, this capability to narrow the focus will become even more important.

Outreach to physicians
While most of this discussion has been on the technical/clinical side, machine learning can also help with sales and marketing efforts by ensuring life sciences organizations are reaching out to the right physicians with the right message at the right time – and in the right way so they can consume it.

For example, physicians in areas with a growing Asian or Hispanic population can be targeted with messages about diabetes screening and treatment since both of those ethnic groups tend to have a higher instance of diabetes.

Essentially, machine learning can help uncover which messages will be most important when to physicians and payers so life sciences organizations can act as a true, value-added partner rather than simply a purveyor of products.

Tremendous promise
Clearly, machine learning holds tremendous promise for life sciences organizations. The key is ensuring they have the mechanisms in place to use all of that data in their models – and the human expertise to understand which discoveries require attention and which can largely be ignored.

While it’s possible to build those systems and expertise in-house, it can be expensive as well as time-consuming. By working with a partner who already has the capabilities in place, life sciences organizations can take advantage of machine learning faster and shorten their time-to-value.

Either way, we’ve only scratched the surface of what machine learning can do for life sciences organizations. It will be exciting to see where it goes next.

Lalithya Yerramilli is Vice President of Analytics at SCIO Health Analytics. She has 15 years of experience in analytics in insurance, healthcare, and life sciences industries working with customer info-base, transactional, physician level, patient level, claims and longitudinal datasets.

© Copyright ASC COMMUNICATIONS 2018. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.

 

Top 40 Articles from the Past 6 Months