Neural Networks for Survival Prediction in Medicine Using Prognostic Factors: A Review and Critical Appraisal

Abstract

Survival analysis deals with the expected duration of time until one or more events of interest occur. Time to the event of interest may be unobserved, a phenomenon commonly known as right censoring, which renders the analysis of these data challenging. Over the years, machine learning algorithms have been developed and adapted to right-censored data. Neural networks have been repeatedly employed to build clinical prediction models in healthcare with a focus on cancer and cardiology. We present the first ever attempt at a large-scale review of survival neural networks (SNNs) with prognostic factors for clinical prediction in medicine. This work provides a comprehensive understanding of the literature (24 studies from 1990 to August 2021, global search in PubMed). Relevant manuscripts are classified as methodological/technical (novel methodology or new theoretical model; 13 studies) or applications (11 studies). We investigate how researchers have used neural networks to fit survival data for prediction. There are two methodological trends: either time is added as part of the input features and a single output node is specified, or multiple output nodes are defined for each time interval. A critical appraisal of model aspects that should be designed and reported more carefully is performed. We identify key characteristics of prediction models (i.e., number of patients/predictors, evaluation measures, calibration), and compare ANN’s predictive performance to the Cox proportional hazards model. The median sample size is 920 patients, and the median number of predictors is 7. Major findings include poor reporting (e.g., regarding missing data, hyperparameters) as well as inaccurate model development/validation. Calibration is neglected in more than half of the studies. Cox models are not developed to their full potential and claims for the performance of SNNs are exaggerated. Light is shed on the current state of art of SNNs in medicine with prognostic factors. Recommendations are made for the reporting of clinical prediction models. Limitations are discussed, and future directions are proposed for researchers who seek to develop existing methodology.

Survival Prediction Landscape: An In-Depth Systematic Literature Review on Activities, Methods, Tools, Diseases, and Databases

“…In Table 1, a comprehensive analysis of the scope of review articles indicates that existing studies can be classified into three distinct groups. I) 9 review papers primarily focus on the application of DL algorithms in survival prediction 47,[49][50][51][52][53][54][55][56] , II) 7 review papers summarise the application of ML algorithms in survival prediction 37,48,[57][58][59][60][61] , and 6 review papers summarise survival prediction methods from three different categories namely statistical, ML, and DL methods [44][45][46][62][63][64] .…”

“…Similarly, Westerlund et al 64 do not explore the potential of multiomics data in terms of cardiovascular diseases. In addition, various review papers completely neglect to address feature engineering in survival prediction 46,47,52,56,57,62 . For instance, Feldner et al 37 despite their focus on dimensionality reduction, fall short in providing a comprehensive summary of current trends in feature engineering approaches with respect to diseases and data modalities.…”

“…For instance, Feldner et al 37 despite their focus on dimensionality reduction, fall short in providing a comprehensive summary of current trends in feature engineering approaches with respect to diseases and data modalities. Futhermore, a small portion of these review papers cover details of few state of the art survival prediction models 49,52,56 . While current review papers summarize survival prediction pipelines partially, there is a necessity to bring diverse information into a unified platform which offers comprehensive insights into patterns and trends associated with survival prediction pipelines.…”

Survival prediction landscape: an in-depth systematic literature review on activities, methods, tools, diseases, and databases

“…In Table 2 , a comprehensive analysis of the scope of review articles indicates that existing studies can be classified into three distinct groups. (I) Nine review papers primarily focus on the application of DL algorithms in survival prediction (Ahmed, 2005 ; Bakasa and Viriri, 2021 ; Kvamme and Borgan, 2021 ; Pobar et al, 2021 ; Kantidakis et al, 2022 ; Altuhaifa et al, 2023 ; Salerno and Li, 2023 ; Wekesa and Kimwele, 2023 ; Wiegrebe et al, 2023 ), (II) seven review papers summarize the application of ML algorithms in survival prediction (Gupta et al, 2018 ; Lee and Lim, 2019 ; Boshier et al, 2022 ; Guan et al, 2022 ; Mo et al, 2022 ; Wissel et al, 2022 ; Feldner-Busztin et al, 2023 ), andsix review papers summarize survival prediction methods from three different categories namely statistical, ML, and DL methods (Bashiri et al, 2017 ; Herrmann et al, 2021 ; Tewarie et al, 2021 ; Westerlund et al, 2021 ; Deepa and Gunavathi, 2022 ; Rahimi et al, 2023 ).…”

“…Similarly, Westerlund et al ( 2021 ) do not explore the potential of multiomics data in terms of cardiovascular diseases. In addition, various review papers completely neglect to address feature engineering in survival prediction (Ahmed, 2005 ; Bashiri et al, 2017 ; Gupta et al, 2018 ; Pobar et al, 2021 ; Kantidakis et al, 2022 ; Rahimi et al, 2023 ). For instance, Feldner-Busztin et al ( 2023 ) despite their focus on dimensionality reduction, fall short in providing a comprehensive summary of current trends in feature engineering approaches with respect to diseases and data modalities.…”