OR WAIT null SECS
Lisa Kerber, Chief Operating Officer at PrognosKerber has more than two decades of experience supporting the Life Sciences industry and specializes in healthcare informatics. Before joining Prognos, Kerber served in various roles at QuintilesIMS (formerly IMS Health) for more than 20 years.
Lisa Kerber outlines how the pharma industry can take advantage of combining new clinical data sources such as diagnostics, or lab data, with AI to deliver on the promise of improving outcomes.
Big data and Artificial Intelligence (AI) undoubtedly have begun disrupting how the business of healthcare is done. As a result, information is available in near real-time as are the business insights derived from it.
For pharmaceutical organizations, this means they can now depend on real-time and forward-looking data versus historical data in decision-making, informing timely actions to benefit the patient.
Today, the pharma industry can take advantage of new clinical data sources such as diagnostics, or lab data, combined with AI to deliver on the promise of improving outcomes. With these new tools, pharma can tailor its efforts to deliver the right therapy to the right patient at the right time. Everyone in healthcare shares this goal of improving outcomes: the pharma industry, payers, physicians, and patients. Tools that allow all stakeholders to align and work toward accomplishing that goal can also enhance collaboration and facilitate creation of strategic partnerships and risk-bearing contracts in the industry.
Traditionally, the pharma industry has used medical claims and prescription data to inform sales and marketing strategy. While valuable, utilization data provides a historical perspective only, helping identify physicians with patients who are already on the relevant therapy, i.e., where a treatment intervention has already occurred. Diagnostic data is available in real time and provides a window into where a treatment intervention is needed, thereby allowing organizations to react to business intelligence in real time and allocate resources in a more targeted way.
Lab data drives about 70% of medical decisions and provides an unrivaled level of specificity for clinical conditions. Standalone lab results (without AI) have a great value by allowing segmentation of patient populations into three groups:
In the case of new patients, lab data allows identification of physicians with relevant patients before a treatment decision is made. The latter two segments (particularly with the addition of data on actual medications prescribed) enables further segmentation of de-identified patient populations based on outcomes and points to future treatment decisions. If a patient is doing well, the current therapy will be continued; however, an intervention is indicated for patients who are not doing well.
On its own, lab data advances organizations one step further in the decision-making process, but adding AI increases the value of lab data exponentially. Coupling billions of lab results with AI enables predictions to be made about which physicians will have relevant patients before a patient is tested and results are delivered. It is precisely this AI capability that holds the greatest value to everyone in healthcare. Getting the patient tested, a procedure done, and the right therapy prescribed sooner can reduce the disease curve. This is particularly useful in conditions that are difficult to diagnose, i.e., the ones that involve a string of misdiagnoses and result in an increased burden for both the patients and the health care system.
When speaking of healthcare AI, it is important to note that accuracy of predictions depends to a great extent on the quality of data used to formulate the predictions. Unlike claims, lab data is not standardized and is often missing critical pieces of information that constitute its value, such as a LOINC code. The first order of business is to use machine learning to predict the code and fill in the missing data. Moreover, oncology lab data may be linked to a pathologist, thereby requiring further work to identify a treating physician. AI can be effectively utilized to increase the quality and value of the underlying data.
Another key element is the amount of data necessary to provide meaningful insights using AI. Identifying patterns to make predictions is only possible with billions of records aggregated from multiple labs. Pharma companies find that collaborating with healthcare AI experts is often the fastest way to get actionable insights, eliminating the need to deal with raw data and build up AI infrastructure.
Once the data is standardized, AI algorithms are applied to find actionable insights. Let us take the example of Rheumatoid Arthritis (RA), a chronic disease with a high-cost burden. While it is generally treated first-line by generics meant to control symptoms, a biologic can go a long way to treat a patient segment that has not seen improvement on the standard-of-care first-line therapies. Identifying and treating these patients with a disease-modifying rather than symptom-controlling therapy will reduce functional disability, decrease pain, improve health-related quality of life and premature mortality - essentially, it will reduce the burden of disease across the spectrum for healthcare systems, patients, employers and the society.
Diagnostic data will allow identifying physicians with new patients, as well as physicians with patients who have been diagnosed and are not doing well on their current therapy. With that information, profiles are created for patients who would benefit from a biologic (or a new biologic) to treat RA.
Tracing these RA patient profiles over time, including those treated by the biologic, and leveraging AI methodologies enables the identification of predictors that can be used to identify those population segments that would benefit most from an earlier or a different biologic prescription. This enables the physician, the patient and the health plan to bypass the costly whirl of treatment trial-and-error and go directly to the therapy that will make a difference.
Thus, intelligence derived from lab data and AI helps pharma organizations do the following:
• Proactively educate healthcare professionals on the benefits of their therapies early in the process, before a patient receives a first-line therapy.
• Reach out to physicians when a first-line therapy fails and the next-line therapy is evaluated.
Lab data has only recently become available for the analytics and AI use in the marketplace. Pharma organizations keenly appreciate the value that high-quality clinical data brings, particularly as the market for companion diagnostics is developing.
At the same time, incorporating such near real-time insights into the sales and marketing strategy requires a change in business operations. Instead of relying on historical data, now an organization needs to incorporate real-time intelligence and educate the sales force to act on it in a timely manner. Technology and big data have accelerated the pace at which decisions are made and require a new level of agility to grow market share.
Augmenting the strategy with a highly targeted outreach can also improve relationships between the industry and prescribers. Real-time data enables a dialogue that happens potentially less frequently but at the right time. Knowing that a physician has a relevant patient now can provide an opportunity for a medical service liaisons to get involved in a particular interaction. As a result, pharma organizations can appropriately allocate their resources while physicians can learn about available, appropriate treatment options before making a decision, which improves and enhances care.
Lisa Kerber is Chief Operating Officer at Prognos.