Skip links

Who gets metastases from a melanoma? An artificial intelligence prediction model (IMPROVE study)

Project summary

Every year, about 7000 cases of melanoma are diagnosed in the Netherlands, the majority of which are in an early stage. As a result, the prognosis is good for many such small melanomas. But it still happens regularly that the cancer has had a chance to metastasize. Of the approximately 800 patients who die from melanoma each year, just under half have been diagnosed with early-stage melanoma.

Effective treatments are available for this high-risk group (such as immunotherapy), but it is not realistic to treat every early-stage melanoma patient with immunotherapy. That would cause massive overtreatment; moreover, immunotherapy is expensive and not completely without side effects. That is why there is an enormous need for a way to determine the risk of metastasis quickly after diagnosis.


The researchers will investigate these melanomas down to the molecular genetic level in order to unravel the secrets of the DNA. In addition, they make digitized microscopic images of all melanomas, so that they can analyze them using artificial intelligence. Combining these microscopic and DNA analyzes with treatment data (how were these patients treated, how did they fare later), creates a wealth of data that can be used to form a prediction model that will benefit new patients with thin melanoma to have.

More detailed information

Principal Investigator:

dr. Loes Hollestein

Role Erasmus MC:




Project website:

Funding Agency: