Table 1 Characteristics of patients in the studies and the quality of the included studies

A Enshaei (2015) [18]

ANN

Northern

NA

OS/R0

ovarian serous

age, stage, grade, histologic type, and preoperative Ca125

Different ML techniques were compared with the data set by examining the models to discover the optimal stop-training point.

AI systems may play an important role in providing predictive data for the treatment and diagnosis of patients.

Low

G Bogani (2018) [19]

ANN

Italy

61.6

R0

secondary cytoreductive surgery (SCS)

NA

ANN analysis was used to weigh the importance of related variables, thus predicting each variable’s impact on achieving CC and survival outcomes.

AI can help in the diagnosis, treatment, and decision-making processes needed.

Unclear

A Laios (2020) [20]

k-NN

UK

64.4

OS

serous ovarian

Age, BMI, Charlson Comorbidity Index, the timing of surgery, surgical complexity, and disease score.

K-NN models were used to classify R0 versus non-R0 patients.

The k-NN algorithm was a promising tool for predicting R0 removal.

Low

A Laios (2021) [21]

ML

UK

64

CCU Admission

advanced stage high-grade serous ovarian cancer (HGSOC)

pre-treatment albumin, surgical complexity score, estimated blood loss, operative time, and bowel resection with stoma

All patient data were used for ANN models in regression mode, and sample selection was varied using the Kennard-Stine method, with 60% for training and 40% for testing.

Predictive ML algorithms may improve the quality of modern care by improving diagnostic prediction accuracy.

Low

A Laios (2022) [23]

XGBoost

UK

63.5

R0

Advanced-Stage Epithelial Ovarian

NA

A new AI-based predictive LOS score was developed for patients with HGSOC after CC. ANN was combined with ordinary logistic regression to predict continuous and binary LOS outcomes for HGSOC patients.

AI can help to predict LOS in EOC patients in the advanced stage after CC.

Low

J Ai (2022) [24]

ML

China

63.5

UTI

ovarian cancer after cytoreductive surgery

age, BMI, catheter, catheter intubation times, blood loss, diabetes and hypoproteinaemia

With the help of ML, five models using two-stage estimation methods of predictor variables were used to predict UTI.

The prediction model based on ML developed using random forest classification can help in treatment decisions, prevent postoperative UTIs, and improve clinical outcomes.

Low

Y Feng (2022) [25]

ML

Netherlands

54.2

OS

ovarian serous

CA125 level, white blood cell (WBC) count, presence of lymph node metastasis (LNM), MO count, the MO/LY ratio, differentiation status, stage, LY%, ascites cytology, and age.

A decision tree algorithm based on ML was used to predict survival.

AI can accurately predict the survival of patients with serous ovarian cancer.

Low

A Laios (2022) [23]

ANN

UK

64

LOS

advanced stage HGSOC

NA

ML and deep learning methods using ANN combined with ordinary logistic regression were used to predict continuous and binary LOS outcomes for HGSOC patients.

Quantitative and qualitative AI models can be highly accurate in predicting LOS in advanced-stage EOC patients after CC.

Low

A Laios (2023) [15]

XGBoost

UK

NA

R0

advanced-stage EOC

NA

An explainable AI framework was used to explain trait effects associated with CC.

AI had adequate precision to explain the effects of the characteristics associated with CC.

Low

S Piedimonte (2023) [26]

ML

USA

58

R0

advanced ovarian cancer (AOC)

CA125, albumin, diaphragmatic disease, age, and ascites

A random forest model was used to predict the optimal CC (< 1 cm) and no gross residual (RD = 0).

The ML algorithm had high accuracy for predicting the optimal cell reduction in patients with AOC selected for PCS, and this method can help in decision-making.

High