Improving public health sustainably with AI in prevention
Perspectives from business, science and clinical practice
Artificial intelligence in prevention marks a fundamental paradigm shift in healthcare: From reactive treatment to proactive health promotion. Through the analysis of large amounts of data, real-time pattern recognition and personalised risk profiles, AI in prevention promises to make interventions not only more efficient but also more personalised. This creates new opportunities to detect disease earlier, target risk factors and influence health behaviour in the long term. But these technological advances also raise complex issues – such as data sovereignty, structural inequalities and the role of human decision-making in interaction with algorithmic systems. We spoke to three experts for the Artificial Intelligence Technology Hub to explore the opportunities and challenges of AI in prevention. Stefan Zipperer is co-founder and CEO of Gesundheit bewegt. Together with his team of prevention experts, he works in the fields of workplace health promotion (WHP), trainee fitness, women’s health, mental health, workplace health management (WHM) and the new trend topic of longevity. The company offers a digital health promotion platform, app-based health campaigns, motivation and gamification concepts, digital health days and prevention AI. Daniel Moll is a digital consultant at DAK-Gesundheit. With more than 5.5 million members, DAK is one of the largest health insurance companies in Germany. As an expert in health promotion, Daniel Moll previously worked as a consultant and project coordinator at the Lower Saxony State Health Association and the Academy for Social Medicine, and develops digital and AI-supported prevention programmes at DAK. University Professor Dr. Andrea Schaller researches and teaches in the field of movement and health sciences, in particular occupational health promotion and medical rehabilitation, at the University of the Federal Armed Forces in Munich and the IST University of Applied Sciences for Management. Her research is based on a social and health sciences perspective and focuses primarily on the individual, but also on organisational structures. Digital tools and artificial intelligence are also playing an increasingly important role in researching the needs-based and target-group-specific design of exercise programmes in different fields of activity and settings. This includes the effectiveness and impact of digital or AI-based physical activity programmes.
What excites you about AI in prevention?
Stefan Zipperer: Prevention is often seen as unsexy and inaccessible. Artificial intelligence can change this – by personalising prevention and translating it into an individual’s life context. This motivates implementation. Our thesis: AI-based prevention can be a real game changer. Particularly as the use of prevention services and general health literacy in Germany are in need of significant improvement.
Daniel Moll: I think AI offers us the opportunity to make prevention much more targeted and individualised. For example, if we have enough data, we can better tailor training or nutrition plans and interventions to people’s needs. Prediction of disease or identification of risk factors can be improved, allowing us to take preventive action earlier.
Prof. Dr. Andrea Schaller: The possibility of low-threshold access to prevention services and the associated opportunity to reach a broad and, in particular, vulnerable target group.
What applications of artificial intelligence in prevention are you researching or using in your organisation?
Stefan Zipperer: We have developed our own health framework that integrates seven key health areas such as diet, exercise and mental fitness. The next step: AI helps us bring together hundreds of tips, evidence-based studies and proven behavioural and motivational models – and derive personalised prevention pathways. Our internal model is called Pepino. We are currently developing a personalised AI-based health coach for healthy work and longevity/50+.
Daniel Moll: For us, the use and development of artificial intelligence in prevention is still in its early stages. The first use cases could be chatbots on prevention and health topics, a kind of health advice service based on evidence-based information that we provide. We are also looking at how we can use existing data to help our members stay healthy. For example, by targeting them in accordance with § 25b SGV V if health risks are identified among the insured. In general, we are looking at appropriate functionalities such as individualisation in new prevention offerings, regardless of whether this is then implemented explicitly by AI or in some other way.
Prof. Dr. Andrea Schaller: We are interested in questions of the best possible integration and synergy creation of “humans and AI”, especially in the context of exercise care. How is the skill profile of exercise providers in therapy, prevention and healthcare changing, taking into account the opportunities and risks of AI applications in exercise delivery?
What are the benefits of artificial intelligence solutions in prevention for patients or policyholders?
Stefan Zipperer: Prevention becomes more effective, more targeted and more personalised. It takes into account personal circumstances and medical history. An example: A smartphone can already detect more than 50 biomarkers. The trick is to bundle this data and translate it into actionable information. This is where LLMs and specially developed health LLMs will come in. They will also become “translators”.
Daniel Moll: With the help of AI, it can be easier to provide policyholders with tailored and individualised solutions. I think policyholders rightly expect to be provided with appropriate preventative services, and AI can help us do that. Another advantage of using chatbots would be that they are available around the clock, not just during business hours. For policyholders, this also means that they can contact their health insurer with health questions at any time, even at the weekend or late at night.
Prof. Dr. Andrea Schaller: In my opinion, the advantage lies in the low-threshold access and the possibility of comprehensive care.
What are the risks of using artificial intelligence in prevention?
Stefan Zipperer: The decision to use and share data lies with the individual. So for us, AI is an opportunity – not primarily a risk. The problem is an overly bureaucratised approach to data protection that limits opportunities and places a general suspicion of misuse on the handling of data. In other words, a sense of proportion and an opportunistic approach. One risk remains, however: People who tend to have fewer socio-economic resources and access to education may not benefit equally from AI-based prevention. Inequality will be exacerbated.
Daniel Moll: When AI is used to predict diseases, the question should always be asked as to what data base the AI was trained on and whether this could lead to biases. I see the same risk when, for example, existing chatbots are used for health issues. It is often unclear to both laypeople and professionals which sources have been consulted to answer questions or provide advice, and how reliable these sources really are. Furthermore, I still do not believe that AI can replace prevention work on the ground and in person. The same hurdles remain as with digital services in general and the risk of increasing health inequalities.
Prof. Dr. Andrea Schaller: I see risks with regard to the quality of (physical activity) care (especially with regard to the conceptual approach of the multidimensionality of physical activity care services), the need for a high level of health literacy among users to deal with the information, the (medium-term) risk of exclusion of vulnerable groups (e.g. through payment barriers) and evidence-based or “monopoly formation” with regard to the information provided by AI.
How will AI affect the labour market in terms of prevention?
Stefan Zipperer: A basic understanding of AI and the use of an AI toolbox will become basic skills – no longer a “nice-to-have”, but necessary in many areas of prevention work. At the same time, the human touch will remain essential: personal contact, empathy and a direct approach will continue to play a central role.
Daniel Moll: I’m not sure that AI can give prevention the big push it needs to have a big impact on the labour market. If hopes are fulfilled and we achieve better disease prevention through AI, this could translate into fewer sick days. I see the greatest potential in mental illnesses and their prevention and possibly treatment with the help of artificial intelligence. But I think there is still a long way to go and we need to focus more on prevention in general.
Prof. Dr. Andrea Schaller: There will be many commercially developed new products and the BGF market will be very dynamic. However, this will not contribute to the urgently needed development of evidence in OHF. The scientific community will and must initially “chase” these developments in order to develop methodological approaches for developing evidence for AI products in WHP. However, as the evidence base in BGF and BGM is currently small, I believe that the additional dynamisation through commercial AI applications also poses great risks for the entire field/health care context. It is already a great challenge to contribute to the development of evidence in workplace health promotion and health management, as these are always complex interventions in complex settings. This makes the unreflected and non-theoretical application of AI in this area even more difficult. I have great concerns about this.
What new skills will health professionals need in the future to use artificial intelligence in prevention?
Stefan Zipperer: There are several skills that play a relevant role. First, understanding and critically classifying the functionality and limitations of modern health AI. Second, the ability to use one’s own AI toolbox with confidence. Thirdly, competence in validating and verifying health content – in other words, in dealing with scientific studies and evidence-based information.
Daniel Moll: I think what is needed above all is an openness to new technologies to engage with them and try them out, and I think this should also be actively encouraged by the company. I think the use of artificial intelligence in our everyday applications, including in the office, will continue to grow and, as has been the case so far, this will be clearly visible in some cases and also partly in the background. The technical expertise required by health insurers to make good use of AI will therefore continue to grow. In the context of health insurance companies, sensitivity to data protection is of course also required with regard to chatbots. Everyone should be aware that personal or health data is particularly worthy of protection and should therefore not simply be entered into any chatbot.
Prof. Dr. Andrea Schaller: In the context of providers of physical activity in prevention, secondary prevention and therapy, this is one of our research topics. Probably more social and methodological competence (in relation to scientific methods and dealing with health information and methods of (sports) education). methods and handling of health information and methods of (sports) pedagogy). Whether health professionals can afford to rely more on AI in terms of professional competence is, in my opinion, questionable.
More information on the topic
The ATLAS contains more articles and science digests on artificial intelligence in prevention. For example, read how intelligent ECG patches combined with deep learning algorithms can help detect atrial fibrillation at an early stage.
You can also read our Bug of the Month interviews on exciting topics.
