Opinion: Data science — the new normal in public health management

Data science has begun to fundamentally change and improve the way many industries operate. This gives those in the global public health and the development sector an opportunity to learn from, adapt, and apply these techniques to address some of the most pressing challenges of our generation.

While data has long been a fundamental tool for public health programs — for example, in epidemiology, monitoring, and evaluation — cloud computing, new data science approaches, and access to large data sets, both health and non-health — can augment current approaches. This has the potential to radically improve the way we manage health systems to achieve greater access and outcomes, giving us better, more sophisticated and timely insights. This can help us to achieve universal health coverage while addressing the increasing burdens of communicable and noncommunicable chronic disease.

First, we can predict and address potential problems before they arise, ensuring we don’t waste precious, finite resources “fighting fires.” Second, new approaches to data use can improve our ability to efficiently and effectively manage health care resources in real-time, to deliver better outcomes. Finally, advances in technology and artificial intelligence allow us to distribute these insights tailored to each individual health care worker across an entire health system daily. This can help them to make better decisions and implement them at scale.

Preventing problems

There are many examples where predictive analytics have transformed industries. The commercial lending industry, for instance, had long relied on traditional measurements such as customer demographics and credit scores to decide loan agreement terms. However, new methods of synthesizing datasets and new technologies that predict default risk, are allowing lenders to assess potential borrowers more holistically.

This allows them not only to extend credit to customers who were previously poorly served, but also to reduce the incidence of loan defaults. If lenders can use new technologies to predict risk of default, why can’t public health leaders do the same to prevent patients defaulting from treatment or even disease outbreaks before they become too costly to fix?

The truth is, they can. BroadReach Consulting recently launched the Vantage-powered Patient Retention Solution — a machine learning algorithm built on an initial dataset of half a million HIV-positive patients on antiretroviral treatment that predicts which patients are at highest risk of defaulting on their next appointment. Appointment defaults are essentially the starting point for patients dropping off treatment. This has significant implications not only for the individual’s health, but for the public health sector more broadly.

Predictive analytics gives health care workers the foresight to intervene early and prevent the problem before it occurs. This can have a number of critical benefits including improved patient health outcomes, a reduction in the cost of replacing a patient who drops off treatment and, in costs associated with expensive second- and third-line treatments and medical complications due to non-treatment; and a reduction in costs associated with controlling new resistant strains of the virus.

To extract the greatest value out of predictive analytics, it is important to consider both the human and technology elements. Therefore, when applying the Patient Retention Solution, BroadReach also streamlines integration with existing workflows and creates simple feedback mechanisms that allow for patient retention interventions to be monitored, evaluated, and improved.

Operational efficiency

Data science and new technologies are driving efficiencies and optimizing operations across diverse industries such as manufacturing, global supply-chain management, financial services, and consumer retail. They offer enormous benefits to public health programs. For example, data science can help public health leaders to benchmark the performance of facilities in their programs. Benchmarking is critical to identifying the unrealized potential in health facilities — a critical trigger for resource reallocation and mobilizing large workforces in the right direction.

Data science is a powerful tool to drive efficiencies in public health and optimize the delivery of patient care. But, while AI can consider a wide variety of factors affecting performance, it cannot always tell you everything that matters. For this, we need human intelligence. It remains critical to also engage health care workers to fully understand their challenges, and work with them to improve performance.

One of the biggest challenges to truly harnessing the power of data science in public health systems is health workers’ resistance to digital transformation. This is often due to a lack of understanding of how data science can be applied in public health to help improve their ability to deliver quality health care at scale. We need to do more to facilitate user adoption and invest in change management so that human and artifical intelligence can work hand in glove.

In order to have large-scale impact in public health, the insights generated from the data science work must be thoughtfully integrated into everyday operations in a way that positively changes health worker behavior, rather than being left to retrospective publications, reports, and arcane data visualizations and dashboards. In short, the integration of data science into public health system management must become the new normal.