Private health insurers increasingly analyze health information automatically. Diagnoses, doctors’ bills, treatment data and claims data can now be recorded in a structured way and compared with each other. Rule-based systems and AI applications make differences visible faster than case handlers who read every bill one by one. This is pointed out in an article in Versicherungsbote from September 28, 2026.
If you want to switch to private health insurance, this shifts a risk. A diagnosis that is missing from the application is less likely to go unnoticed when you make a claim. This makes what happens before the application all the more important.
What insurers can compare automatically
When you make a claim, the insurer receives doctors’ bills with diagnosis codes, hospital reports and prescriptions. This information can be compared with the answers in your original application. Through the medical confidentiality waiver, the insurer can also request information from the doctors who treated you.
A typical example: two years after the start of your contract, you submit a bill for orthopedic treatment. The bill shows a back diagnosis that, according to the doctor’s report, has existed for years. It was not stated in the application. A system can flag this difference automatically, and the insurer then asks for documents.
What is new is not that insurers check such differences. What is new is how fast and how completely this can happen.
A flag is not yet a breach of the duty of disclosure
When software reports a difference, this does not yet mean that you did something wrong in the application. Under § 19 VVG, you only have to state the risk factors that you know about and that the insurer asked about in text form. Whether there is a breach therefore depends on several points:
| Point to check | Why it matters |
|---|---|
| Specific health question in the application | You only have to state what you were asked. A diagnosis that no question covers is not a breach of duty. |
| Period covered by the question | Many questions refer to the last three, five or ten years. Anything before that was often not asked at all. |
| Actual knowledge | You must have known about the diagnosis. Something the doctor billed but never told you about is assessed differently. |
| Medical facts | A one-off muscle tension is different from a treated chronic illness, even if both can carry the same code. |
| Course of treatment | Whether a complaint had ended or was still being treated decides how much it matters for the application. |
So the technical flag is the start of a review, not its result. In the end, every case has to be assessed legally and medically on its own.
The insurer is also limited in time. Under § 21 (3) VVG, its rights expire five years after the contract was concluded, or after ten years in the case of an intentional or fraudulent breach. However, this limit does not apply to insured events that occur within this period.
Where the real risk lies
The danger rarely comes from diagnoses you left out on purpose. Much more often, it comes from entries you knew nothing about or never saw as an illness:
| Type of entry | How it comes about |
|---|---|
| Wrong or unclear billing diagnosis | A doctor codes a diagnosis so that a service can be billed. It has little to do with your actual findings. |
| Suspected diagnosis | A suspicion is documented but later never confirmed or explicitly ruled out. |
| Diagnosis code carried forward | A complaint that healed long ago keeps appearing in the bills quarter after quarter. |
| Mental health diagnosis | A depressive episode or adjustment disorder is billed, for example in connection with a sick note, without you learning about it. |
These are exactly the entries that an automated comparison finds reliably. And with exactly these entries, it is hard in a dispute to prove later what the situation really was at the time.
Why hoping is no longer a strategy
In consultations, we hear a sentence like this again and again: “I just hope that nobody finds this diagnosis later.” That was never a good idea. With structured data analysis, it becomes a bad strategy.
If you conceal a diagnosis that you knew about and were asked about, you risk much more than a rejected bill when you make a claim. Depending on the degree of fault, the consequences range from a later risk surcharge or benefit exclusion to rescission of the contract. Anyone left without cover after several years in private health insurance often cannot get a new contract on comparable terms.
How to prepare your health history
The better answer to automated checks is an application that stands up to any check. From over 5,000 applications with health questions, we know that most problems can be solved before the application, not afterwards.
- Check your patient records. Ask your doctors for access or a copy and read the diagnoses for the years the questions cover.
- Request a statement of services from your statutory health insurance fund. Under § 305 SGB V, your statutory health insurance fund must tell you on request which services you have used. It also shows the diagnoses that were billed.
- Clarify unclear diagnoses with your doctor. If a diagnosis is wrong, your doctor can confirm this. With this proof, you can ask your statutory health insurance fund to correct it.
- Use medical statements. A short confirmation that a complaint has healed or that a suspicion was not confirmed settles many questions.
- Document your answers in a traceable way. Write down which documents you used for your answers. This helps if questions come up years later.
- Make an anonymous risk enquiry. Only with a prepared history is the anonymous risk enquiry worthwhile. The insurers then assess the complete picture, and a rejected application does not end up in your file.
We explain in detail what consequences a false answer can have on our page about the pre-contractual duty of disclosure.
Our assessment
AI does not change the rules of the duty of disclosure. It changes the likelihood that gaps are noticed. For honest applicants, this is no disadvantage as long as their documents are in order. The risk lies mainly with people whose records contain diagnoses they know nothing about.
Our recommendation is therefore clear: first clarify your own health history, then make an enquiry, then switch to private health insurance. This order costs a few weeks, but in a serious case it saves you the dispute about what was on a bill years ago.
Frequently asked questions
May my insurer analyze my health data with AI?
Insurers process the health data they receive from you in the application and when you make a claim. How they analyze this data technically, whether with classic case handling, fixed rules or AI, does not change the legal standards of the duty of disclosure. What counts is still what the application asked and what you knew at the time.
How long can the insurer rely on a false answer in the application?
Under § 21 (3) VVG, the insurer's rights expire five years after the contract was concluded. If the duty of disclosure was breached intentionally or fraudulently, the period is ten years. This limit does not apply to insured events that occurred within these periods.
What can I do if my data at the statutory health insurance fund contains a wrong diagnosis?
You can ask your statutory health insurance fund to correct it. Under § 305 SGB V, you need a medical certificate showing that the diagnosis is incorrect. The fund must decide on your request within four weeks. Clarify this before you apply for private health insurance, not afterwards.
