ESMO Congress 2024 | OncologyPRO

Abstract 2013P

Background

Decisions on the best available treatment in clinical oncology are based on expert opinions in multidisciplinary cancer conferences (MCC). For this purpose, artificial intelligence (AI) could increase evidence-based treatment in clinical oncology as assistance system to give an additional treatment recommendation in MCC. Here, we present the first preliminary data for urothelial carcinoma (UC) patients AI-generated treatment recommendations.

Methods

We have transformed comprehensive patient data (107 individual features) of 1029 MCCs for UC from the years 2015 - 2022 into representations that can be used in software development. Next, we developed a two-step process in order to train a classifier to mimic the MMC recommendations. In the first step, we identified superordinate categories of the recommendations. In the second step, we specified the detailed recommendation. For this purpose, we used different machine learning (CatBoost, XGBoost, Random Forest) and deep learning (TabPFN, TabNet) approaches. Accuracy weights are determined by F1-Score.

Results

We developed an AI system which is able to decide which kind of superordinate recommendation should be applied, e.g. surgery or anticancer-drugs (Table). Furthermore, our AI system is able to suggest the specific surgical treatment as well as the correct drugs. Finally, the results show that our selected deep neural network architectures are able to learn from the limited amount of data. Table: 2013P

Accuracy rates for AI-generated treatment recommendations of urothelial cancer based on F1-Scores

Task	F1-Score ↑	#Classes	Class	F1-Score
1st Step	0.7912	5	Surgery Medication Aftercare Chemoradiotherapy Best supportive care	0.8831 0.8842 0.5714 0.0 0.6667
2nd Step: Surgical Prediction	0.6500	5	Cystectomy Cystoprostatectomy TURBT Nephrectomy Nephroureterectomy	0.7660 0.5 0.5714 0.0 0.6667
3rd Step: Drug Prediction	0.6880	12	Gemcitabine/Cisplatin Gemcitabine/Carboplatin Vinflunine Paclitaxel Avelumab Pembrolizumab BCG Mitomycin Carboplatin Nab-Paclitaxel Paclitaxel/Gemcitabine Enfortumab-Vedotin	0.9804 0.5 0.0 0.0 0.0 0.6667 0.6667 0.0 0.0 1.0 0.0 0.6667

Conclusions

To our knowledge, we present the first time data for fully automated AI-based treatment recommendations for MMC in urothelial cancer with excellent accuracy rates. In future, we aim to implement clinical trial data to enable an explainable AI for the generated recommendations.