June 2023. GrowthPolicy’s Devjani Roy interviewed Soroush Saghafian, Associate Professor of Public Policy at Harvard Kennedy School and founder/director of the Public Impact Analytics Science Lab (PIAS-Lab) at Harvard, on data analysis and smart policy design, transformational health care, and mitigating risks in AI. | Click here for more interviews like this one.
Links: Faculty page | Public Impact Analytics Science Lab (PIAS-Lab) at Harvard
GrowthPolicy: You are the founder and director of the Public Impact Analytics Science Lab (PIAS-Lab) at Harvard. Please tell our readers a little bit about the origin story of the PIAS-Lab and the vision that led to its founding. Also, if you’d like to share with our readers any of the PIAS-Lab’s important projects and achievements, past or present.
Soroush Saghafian: In my lab, the Public Impact Analytics Science Lab (PIAS-Lab), we are collaborating with a variety of organizations to solve problems that can have public impact. The PIAS-Lab is devoted to advancing and applying the science of analytics for solving societal problems that can have public impact. We take a problem-driven approach, meaning that we make use of the best analytics science methods to address each unique problem most effectively. These tools come from various branches of analytics science, including operations research, machine learning and big data, decision science, statistics, and artificial intelligence, among others.
We have been using these tools to help hospitals, startups, public agencies in the United States and beyond, and private firms solving problems that have public impact. The tools and related collaborations with these entities have enabled us to find the best ways possible to save lives, improve the quality of care delivered to patients, decrease healthcare expenditures, reduce existing inequalities, design superior policies, and make better use of technological advancements such as mobile health, smart devices, and telemedicine, among others.
GrowthPolicy: One of your many areas of expertise is healthcare policy and operations management. How have the healthcare provisions in the Biden Administration’s Inflation Reduction Act (December 2022) impacted access to healthcare and prescription drug prices? What meaningful changes can policymakers pursue to improve the affordability and quality of healthcare in the United States in the aftermath of the COVID-19 pandemic?
Soroush Saghafian: Each year the U.S. spends about $4 trillion in its healthcare sector. Per capita, that is over 2.5 times of the average spending across other developed countries. Yet, in terms of quality of care and patient outcomes, we are not comparable with those countries.
The healthcare sector in the U.S. is incredibly complex, with a vast variety of problems. Obviously, a simple act like the Inflation Reduction Act (IRA) of 2022 will not resolve many of these problems. Having said that, part of IRA was designed to provide more power to Medicare to negotiate some drug prices. Most optimistic estimates show that, if successful, IRA may result in a $300 billion reduction over the next ten years, which is tiny compared to the size of some of the problems in the U.S. healthcare sector. Nonetheless, IRA has some good intentions, and it is in the right direction. For starters, Congress was finally able to put some limit on the incredible power of the pharmaceutical industry. IRA also increases eligibility for low-income subsidies and enhances access to a few drugs. But, again, a single act like IRA, which is not even a healthcare act per se, will not be able to resolve the extraordinary number of problems faced in U.S. healthcare.
To improve healthcare, we need careful data analysis (not just good intentions). Evidence obtained from careful data analysis can, for example, help us move away from traditional delivery and payment systems. But good intentions are not enough, especially in dealing with the COVID-19 pandemic and its aftermath.
COVID-19 put an enormous financial and operational stress on the U.S. healthcare system. In response, the government provided stimulus packages such as the Coronavirus Aid, Relief, and Economic Security (CARES) Act, which earmarked $175 billion for healthcare providers hit hard by the pandemic. The Department of Health and Human Services also expanded the Medicare Accelerated and Advance Payment program—a loan program that helps hospitals with disruptions in cash flow. These were well-intended policies that lacked careful data analysis. The money provided by the CARES Act, for example, disproportionately went to wealthy hospitals. The lesson is that we need to make use of data and carefully designed analytical models and algorithms in designing healthcare policies, especially if we want to learn how limited resources should be allocated to address major problems.
We also need to make use of these tools to move away from traditional delivery methods that require patients to physically go to hospitals or doctor offices. Mobile Health (mHealth) and Hospital-at-Home based technologies are ripe to make use of the power of AI and smart devices to reduce barriers in access to care. In parallel, we need to also move away from traditional payment mechanisms such as volume-based payments that introduce wrong incentives, and, instead, create a value-based payment system, enabling providers to focus on quality.
Finally, we need to substantially improve transparency of quality outcomes across hospitals and providers, and think of innovations that allow policymakers to efficiently and publicly report measures of quality. Patients should be empowered to better align their needs with the capabilities of high-quality providers, if we want to have a higher-quality healthcare system.
GrowthPolicy: I’d like to ask about your view on regulatory mechanisms for AI. As AI and machine learning are becoming more popular, what steps can be taken to ensure that these tools are used in a responsible manner?
Soroush Saghafian: Imposing appropriate regulations is important. There is, however, a fine line: while new tools and advancements should be guarded against misuses, the regulations should not block these tools from reaching their full potential.
I have emphasized many times that the possibilities in using AI along with the available data in creating positive impact are beyond limits. But like many other technological advancements, advanced analytics science tools can be used in a positive or negative way. Naturally the main public fears about them are around potential misuses. However, the possibilities for positive uses and solving societal problems are also vast.
This is partly because policy decisions are naturally complex and extremely challenging. The AI and machine-learning branches of analytics science are great tools, because they allow us to move away from opinion-based solutions and, instead, adopt data-driven strategies. To harness them responsibly though, we must make use of them in specific ways. For example, we need to make sure they are not solely trained on data generated by human decision-makers, which, by nature, are often biased towards their own views.
GrowthPolicy: You wrote about the opioid epidemic in a recent research paper, “Understanding the Opioid Epidemic: Human-Based Versus Algorithmic-Based Perceptions, Treatments, and Guidelines.” What are the potential benefits of an algorithmic-based approach to managing opioid prescriptions? And, a follow-up question: how can policymakers and individual physicians utilize your research insights to address the current opioid epidemic?
Soroush Saghafian: The opioid epidemic caused over 556,000 deaths in the U.S. between 2000 and 2020, and yet there is still no complete understanding of what can be done to effectively control this deadly epidemic. We have developed advanced algorithms trained on large-scale data that can help improve opioid-related policies, which have shown superior recommendations compared to the Centers for Disease Control and Prevention’s (CDC’s) 2016 and 2022 guidelines.
This is just one example of how we can harness the power of algorithms and large-scale data to save lives and improve society. But it requires policymakers and other authorities to make use of algorithm-driven solutions and recommendations, or, at least, involve them in their decision-making process. I do not want to imply that policymakers or other authorities are always against these algorithm-driven solutions. There is some eagerness to make use of advanced algorithm-driven solutions. However, some policymakers still prefer traditional decision-making processes and end up ignoring what carefully designed algorithms can offer.
In summary, while the potential for creating positive public impact through such carefully designed algorithms is endless, they will not reach their full potential unless policymakers and other authorities that make large-scale decisions realize the fact that, in many domains, they need to move away from traditional decision-making methods and adapt to the age of AI.