New data era for hospitals

Hospital reform is coming. In a much more complex billing world in the future, there will be a high demand for data analysis and predictive simulation. Hospitals should not put this on the back burner, says Dr André Sander, head of technical development at ID, a company specialising in the analysis of digital hospital data. On the clinical side, this also applies to AI applications that can provide comprehensive support to doctors - if done right.

Artificial intelligence is a hot topic for digital enthusiasts around the world. You are giving a presentation on AI in medical decision support at the DMEA. What is really needed - and what is just marketing?

What doctors and other healthcare professionals want from AI is to make their jobs easier. AI that is truly useful does just that. It's not about taking decisions away from a doctor, or even anticipating decisions. It's about support, it's about efficiency, and ultimately, it's about economics. From our point of view, AI support for diagnosis is not the most important thing. We are primarily looking for applications that, in addition to decision support, address other aspects of care, particularly in the documentation and further processing of clinical data. For us at ID, the use of AI here is the core of artificial intelligence. And this is what we are demonstrating at the DMEA.

What are the challenges in implementing 'smart' documentation?

The challenge is always to integrate the relevant applications into everyday workflows. No one wants to have to open a new application or log in to a new place just to use AI. It must be seamlessly integrated into the treatment context. A second challenge is that AI must be completely transparent and understandable, especially when it comes to medical decisions. Users do not want to have to rely on a large language model (LLM) that sometimes hallucinates and must be looked up every time to check that the answer is correct.

How is ID addressing these challenges? And what can we see of this at the DMEA?

In terms of AI models, we combine traditional rule-based AI with statistical AI, and we combine business intelligence approaches with data-driven approaches. The latter have the advantage of being easy to simulate. Combined with knowledge-based modules, they can be used to create solutions that are easy to understand. One application we will be demonstrating at DMEA is predicting length of stay based on clinical documentation. This has both an economic and a medical dimension.

What about medical device certification for this and similar applications?

We are very much in favour of certifying everything as a medical device, at least Class IIa. If I were running a clinic, I would be very wary of software products that are only Class I, to put it mildly. As soon as software interferes with the treatment process, it must be Class IIa anyway. Another regulatory issue we are increasingly dealing with is the AI Act. It is not easy to get through. In the future we will have to prove exactly where the data we use to train AI comes from, and that we will also have to assess bias. Anyone who, like us, has already relied on Class IIa and higher medical device certification has already established some of what is required for the AI Act. However, the AI Act goes beyond the requirements of the Medical Device Directive (MDR) in some respects. The MDR is essentially about risk control. The AI Act is also concerned with the quality of outcomes.

The federal government is still in the process of reforming the hospital system. How will this change the way hospital management teams work with their data?

Hospitals will increasingly have to deal with structural data, and so will we as IT and data service providers. How is a hospital organised structurally? Will it be able to serve certain service groups in the future or not? Answers to these questions are needed, and they are not easy to find. We will also be increasingly concerned with the integration of outpatient and inpatient care. After all, at least one of the aims of hospital reform is to reduce the number of beds and replace some inpatient care with outpatient or at least hybrid care. To make sensible decisions as a hospital and to optimise patient management, it must be possible to compare outpatient and inpatient care scenarios. Simulation of different scenarios will play an important role. Hospitals should start preparing for hospital reform today, not tomorrow.

What does this mean for ID as a company?

For us, this means, among other things, that in the future we will no longer act solely as a software provider but will also support the transformation of hospitals in a much more action-oriented way through targeted analyses. To this end, we have entered a strategic partnership with MEDIQON GmbH, the leading provider in the field of performance planning and management. Together, we will offer a solution that supports hospitals in meeting the requirements of the hospital reform by combining our expertise in hospital controlling and portfolio and care analysis. We will provide comprehensive information on the details at the DMEA. MEDIQON is a company that focuses on planning and simulation tools using interactive data analysis in the healthcare sector. This makes recommendations for portfolio management directly tangible. This complements our services very well.

Last but not least, ID specialises in hospital billing. What about the related IT applications? To what extent will they change because of the hospital reform?

This is already a big challenge. We must map all payment models in all areas, i.e. DRGs, flat rates, hybrid DRGs and, if necessary, normal outpatient billing. Let me put it this way: if one developer built something in the past, there will be three in the future. Nothing will be cancelled. The DRG world in all its complexity will remain, even if DRGs account for only 40% of total revenue in the future.

Sounds like a golden age for controlling.

That's right, yes. Keyword controlling: A controlling tool that we also want to highlight at the DMEA is our ID EFIX® PHARMA. This is an application that makes pharmaceutical data available for controlling. This didn't exist before because only a few hospitals had digital medication. But that is changing. On the one hand, ID EFIX® PHARMA has an economic dimension: we can use it, for example, to find any high-priced medication that could possibly be billed separately. But the application also has a medical dimension, because it provides valuable data for medication therapy safety (AMTS). We can use it to provide comprehensive information on contraindications, interactions or incorrect dosages. This can be of great interest to many hospitals.