Can AI used by your doctor reduce your insurance premium

Health insurance has become one of the most expensive insurances for private households across Europe, with Switzerland leading the statistics at €7000 per capita in annual health insurance premiums. A proper health coverage has become not only expensive, but even unattainable for parts of the European population. In light of the current momentum around AI technology, I recently found myself in a lively debate whether health insurance providers could offer discounts for patients who go to doctors who use AI.

AI in healthcare is clearly a game changer, and applications are no longer a matter of choice, it has become a necessity. AI can help with early diagnosis and personalized treatment plans. It help doctors to reduce healthcare costs.

It appears a rather reasonable question if insurers should consider discounts for doctors who use AI-assisted treatments. AI augmented health treatments are not new, and they can achieve both, lower costs and more personalized and better healthcare outcomes. The assumption is that clinical AI improves the quality in diagnosis and treatment. Only a few months ago, a study by MIT Sloan and BCG found that AI could decrease cost of care by up to 16%.

1. Do we need AI-assisted doctors? One question, two answers

At an recent industry event, a small group of insurance enthusiasts engaged in a engaging discussion over this pressing question. While a single break as not enough for us to get to the bottom of the topic, we did come to one key conclusion:

While it seems important to consider the potential of AI-assisted medical treatments, the more powerful lever of AI applications is through prevention measures on the policyholders' side. To reduce costs, mitigating the risks of costly late-stage therapies or, ideally, falling sick altogether is critical. Utilizing AI to drive behavioral measures to sustain our wellbeing is one of, if not the, most desireable approach.

2. First Response: The most effective way to leverage AI for reducing healthcare costs is through prevention measures for policyholders themselves.

Yet here is the issue: insurers have consistently failed to establish economically viable prevention models at scale. And this is partly due to all of us - our "bad habits". Many people are hesitant or unwilling to adopt new habits early enough to prevent the most common chronic diseases. Additionally, those who do adopt daily tracking devices and sensors to improve their wellbeing are often the ones who need it the least.

To make matters worse, unhealthy behaviors have literally become lethal. Obesity has moved up the ranks to become one of the most common causes of death (WHO) in Europe, largely due to unhealthy dietary habits and a lack of physical activity. This highlights the need to the (behavioral) root causes of rising healthcare costs.

Insurers have struggled to establish economically viable prevention models at scale, and this is partly due to - all of us, the user behavior. Many people are hesitant or unwilling to adopt new habits to prevent common chronic diseases.

Of course, you cannot simply blame it on "user". May European healthcare systems suffer from a lack of cooperation among the main actors, hospitals continue to be incentivized to prioritize profit over prevention, and the pharmaceutical industry's maintains powerful interests in increasing revenues with frequent medication.

As a consequence, the question remains unresolved: While prevention measures driven by users are not yet widespread enough to significantly reduce healthcare costs, should we incentivize practicioners and telecare centers to adopt AI-assisted methods as a quicker pathway to cost reduction?

3. Second Response: While AI-augmented medical practices will improve healthcare outcomes and reduce costs, it remains insufficient to warrant lower health insurance premiums.

While AI-augmented practices used by doctors can improve healthcare outcomes and generate cost savings, it remains insufficient to warrant lower premiums. In fact, insurers should not give insurance discounts for doctors who use AI-assisted treatments.

Here is why: health insurance premiums are based on factors, such as age, medical history, and lifestyle habits. The industry would need to develop a model to price in potential benefits and risks of AI-augmented practices and consider a variety of other risk factors (national health interests, data protection, alternative treatment options) before deciding on any premium discounts.

Put differently, any decision in insurance needs to be based on sound actuarial principles. "Pricing" in AI-augmented practices means that actuaries develop an critical understanding of latest AI science. This comes across as rather complex, and it remains highly questionable if it justifies the return:

• Benefits of AI-augmented practices have not yet been realized broadly and in main areas of treatment. It is simply too early to be understood how to correlate premiums and actual costs of giving "AI-powered care".
• AI algorithms can be susceptible to bias, overreliance and inequity when incentivizing AI-use. Doctors are not immune to bias either, however the potential for unequal treatment of patients, with some receiving AI-care and others not, raises significant ethical implications. Practicioners may end up prescribing a different treatment to a patient on a cheaper, AI-funded scheme. You may argue that this triage is already happening today in private/public health systems, but this is not what we want if we strictly aim for superior care outcomes.

4. Preferred: health insurance discount if you suscribe to AI-assisted prevention.

It appears far too early and risky for insurers to give insurance discounts for doctors who use AI-assisted treatments.

To move forward, and as a preferred option, insurers could incentivize policyholders to use AI-based applications for prevention, which includes early detection tools, self-checks, and monitoring devices. Already today, AI-based remote applications can identify skin cancer as accurately as a group of dermatologists, and some countries mandate the use of early detection tools before seeing a practitioner.

Even though there is a question mark around scalable user adoption and sustainable usage over time, AI-augmented prevention is addressing the root causes of the cost of care, in an action-driven way. (Personally, I would happily cut down on my health insurance premiums, keeping ownership of my health data (e.g. blockchain).)

Alongside, we concluded in our debate, there are two additional measures to consider:

• Clear guidelines for how providers use AI in healthcare, aligned with the ethical principles of medicine. These guidelines can ensure diversity in datasets, testing for bias, and provider training - or certification - in the ethical use of AI.
• Incentivize doctors on health outcomes - rather than AI use. Good outcomes can be reduced hospital re/admissions or improve patient satisfaction scores. I recently learned that doctors who use AI-assisted micro-surgery reduce hospital stays by 10-12 days for specific interventions. This translates into an invaluable recovery upside for patients and significant cost saving for all of us, the payers.

What are your thoughts about this debate?

Mirjam Bamberger is member of the Management Committee of 安盛 's European Markets & Health. Prior to this she has been she has been?CEO of AXA Luxembourg and CEO of AXA Wealth Europe and served in various roles as a board member of AXA Switzerland. Mirjam spent over 20 years living and working in the US and UK, China, Latin America and Europe, holds an honors degree of the executive MBA of IMD Lausanne and a master’s degree of University of Cologne. She is a certified director of the Swiss Board School.