Skip to main content
. 2022 Oct;11(10):3853–3868. doi: 10.21037/tcr-22-1626

Table 3. Summary of identified model-agnostic XAI studies in oncology.

Study Cancer type XAI Framework Purpose of XAI
Delineation of global feature importance Characterization of individual prediction feature importance Visualization of nonlinear relationships and interactions Identification of prognostic and/or predictive thresholds Feature selection
Prognostication
   Li et al., 2020 (15) Prostate SHAP
   Bertsimas et al., 2021 (16) Pancreatic SHAP
   Jansen et al., 2020 (17) Breast SHAP, LIME
   Chen et al., 2021 (18) Nasopharyngeal SHAP
   Moncada-Torres et al., 2021 (19) Breast SHAP
   van den Bosch et al., 2021 (20) Colorectal SHAP
   Gong et al., 2021 (21) Esophageal SHAP
   Farswan et al., 2021 (22) Multiple myeloma SHAP
   Alsinglawi et al., 2022 (23) Lung SHAP
   Jacobson et al., 2022 (24) Bone metastases SHAP
Diagnosis
   Suh et al., 2020 (25) Prostate SHAP
   Kwong et al., 2022 (26) Prostate SHAP
Radiomics
   Manikis et al., 2021 (31) Glioma SHAP, LIME
   Gaur et al., 2022 (32) Brain tumor SHAP, LIME
   Lee et al., 2021 (33) Breast LIME
   Kha et al., 2021 (34) Glioma SHAP
   Du et al., 2019 (35) Nasopharyngeal SHAP
   Wang et al., 2022 (36) Breast SHAP
Pathology
   Palatnik de Sousa et al., 2019 (37) General LIME
   Abir et al., 2022 (38) Leukemia LIME
   Modhukar et al., 2021 (39) Breast LIME
   Chakraborty et al., 2021 (40) Breast SHAP, LIME
   Janssen et al., 2022 (41) Prostate SHAP
   Meena et al., 2022 (42) Skin SHAP
   Park et al., 2021 (43) Leukemia SHAP
Treatment selection
   Ladbury et al., 2022 (44) Endometrial SHAP
   Zarinshenas et al., 2022 (45) Lung SHAP
   Laios et al., 2022 (50) Ovarian SHAP
   Bang et al., 2021 (51) Gastric SHAP
Epidemiology
   Ahmed et al., 2021 (52) Lung SHAP
   Kobylińska et al., 2022 (53) Lung SHAP
   Richter et al., 2019 (54) Melanoma SHAP
Radiation treatment workflow
   Siciarz et al., 2021 (55) Brain tumor SHAP
   Chen et al., 2022 (56) Prostate SHAP, LIME

XAI, explainable artificial intelligence; SHAP, SHapley Additive exPlanations; LIME, Local Interpretable Model-agnostic Explanations.