Supporting hospitals in times of change

In that context the digital data trove is an asset with heightened potential, says Joachim Meyer zu Wendischhoff, head of Medicine and Product Management at ID, which specialises in analysing digital hospital data records The is about far more than helping to code invoices.

Everyone is talking about AI. Your company has been analysing data in the hospital environment for decades now. Do you talk about AI too?

We talk about AI too, in fact about two kinds of AI. We have long been using rules-based AI for coding. Whenever certain terms appear in documentation, an ICD or OPS code is suggested. Certain formulations do not trigger such an action, for example when a diagnosis list says ‘condition after myocardial infarction’ instead of just ‘myocardial infarction’. We are increasingly using machine learning (ML) to support this rules-based analysis. For example, we provide an ML-based module which automatically scans medical reports and generates suggestions for coding an invoice. This has decisive advantages. Manual input is still required all the same, for without coding specialists there would be no chance of success.

How do you train this tool? Is interoperability among hospitals assured?

We deal with everything presented to us in text form: registration reports, vaccination status, surgery reports, discharge reports, laboratory reports, consultation, x-ray reports, as well as progress documentation. This is all analysed, and the algorithm learns which constellations most frequently contain which codes. In theory, interoperability among hospitals is assured. However, this kind of documentation contains what one could term “hospital speak“, which we have to take into account. There are basic components which we use in every hospital, but adapting to the specific documentation in each case is based on individual documents. Sometimes manual input is necessary. There is what is known as a billroth procedure for stomach operations. If by coincidence a doctor in the hospital is called Billroth, then naturally that cannot be the suggestion for a corresponding code each time.

As well as AI and coding, what other topics will ID focus on at DMEA?

The Hospital Reform Act will put a considerable strain on hospitals, and this is where our solutions can help. The reform specifies that every case is assigned to a particular service group and that hospitals receive a fixed allowance for making services available for said groups. At the same time these service groups are intended to provide a template for hospital organisation. We address both aspects. On the one hand we have designed our own classification tool for service groups, based on existing service groups in North Rhine-Westphalia. We also have an extra module, integrated in Kompass, which we co-developed with our partner MEDIQON. Hospitals can use this to analyse in detail what their competitors are doing. This will be a vital topic in order to survive in the years to come. In the case of smaller hospitals the issue is whether certain services can still be provided. Larger hospitals must anticipate taking on extra patients from the outskirts and organise their availability accordingly.

These service groups are not the only innovation. What about hybrid DRGs?

We have already covered those with ID DIACOS®, and they are already contained in the case allowance catalogue. This is not such a big issue for hospitals. The classification tool handles that too. Independent of hybrid DRGs, there is basically a strong tendency towards outpatient treatment, an area where we support hospitals. We have developed coding rules for outpatient care and an EBM search for example. Our main goal is to support hospitals, whatever challenges they face. One of the things we want to talk about at DMEA is the growing interlocking of our modules. We have combined our ID MEDICS® medication tool with the ID EFIX® financial controlling tool. This enables an analysis of medication costs by case and also makes comparisons regarding the efficicacy of drugs as well as analysing prescriptions, e.g. in the case of antibiotics. Another first at DMEA where medication is concerned is our insulin plan, which provides significantly better documentation and treatment for diabetes patients. The whole issue surrounding ePrescriptions is another hot topic in hospitals. They are not mandatory yet for inpatients, but we are already marketing a solution.

Does your medication tool come into contact with electronic patient records (ePRs)?

Of course. This is primarily an issue concerning standardisation and interoperability. We can use ID MEDICS® FHIR to export data records, and we have good knowledge of ISIK conventions. We have long been providing support for the electronic medication plan. In short, this is all geared towards providing input for electronic patient records as soon as it becomes possible. On the subject of interoperability, our ID LOGIK® TS terminology server is naturally important too. Our philosophy is not to dictate terminology to users but to give them the option of converting it if required, for example ICD-10 invoice codes into SNOMED CT codes, which are important for ePRs and other applications. This is also relevant for research, a field where our terminology server is involved at the SMITH consortium of the Medical Informatics Initiative.

What other news is there?

There are a number of fascinating research fields which have not yet evolved into products, which we will also be talking about at DMEA. Together with the Charité hospital and as part of the dVP_FAM project, we are developing an app for breast cancer patients which enables outpatient units, patients and resident practitioners to provide data input. This is an intersectoral tool which has made a promising start, and we are very satisfied. We want to pioneer intersectoral approaches because we consider this a forward-looking topic.