An Assessment of the Predictive Performance of Current Machine Learning–Based Breast Cancer Risk Prediction Models: Systematic Review

Gao, Ying; Li, Shu; Jin, Yujing; Zhou, Lengxiao; Sun, Shaomei; Xu, Xiaoqian; Li, Shuqian; Yang, Hongxi; Zhang, Qing; Wang, Yaogang

doi:10.2196/35750

Cited by 7 publications

(4 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…AI techniques range in complexity from machine learning (ML) algorithms to deep learning (DL). While simple models have been used since the 1960s in the form of logistic regression, recent decades have seen the development of sophisticated neural network algorithms to predict risk for breast, lung, and other cancers [20][21][22][23][24][25][26][27]. The area under the curve (AUC) is a commonly used metric in risk modeling.…”

Section: Artificial Intelligencementioning

confidence: 99%

Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma

Jiang

Chao

Culp

et al. 2023

Cancers

View full text Add to dashboard Cite

Pancreatic cancer is projected to become the second leading cause of cancer-related mortality in the United States by 2030. This is in part due to the paucity of reliable screening and diagnostic options for early detection. Amongst known pre-malignant pancreatic lesions, pancreatic intraepithelial neoplasia (PanIN) and intraductal papillary mucinous neoplasms (IPMNs) are the most prevalent. The current standard of care for the diagnosis and classification of pancreatic cystic lesions (PCLs) involves cross-sectional imaging studies and endoscopic ultrasound (EUS) and, when indicated, EUS-guided fine needle aspiration and cyst fluid analysis. However, this is suboptimal for the identification and risk stratification of PCLs, with accuracy of only 65–75% for detecting mucinous PCLs. Artificial intelligence (AI) is a promising tool that has been applied to improve accuracy in screening for solid tumors, including breast, lung, cervical, and colon cancer. More recently, it has shown promise in diagnosing pancreatic cancer by identifying high-risk populations, risk-stratifying premalignant lesions, and predicting the progression of IPMNs to adenocarcinoma. This review summarizes the available literature on artificial intelligence in the screening and prognostication of precancerous lesions in the pancreas, and streamlining the diagnosis of pancreatic cancer.

show abstract

Section: Artificial Intelligencementioning

confidence: 99%

Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma

Jiang

Chao

Culp

et al. 2023

Cancers

View full text Add to dashboard Cite

show abstract

“…However, as with prediction models as a whole, machine learning tools are also at high risk of bias, as revealed by the PROBLAST tool. (see 9 for review).…”

Section: The Current and Evolving Landscape Of Breast Cancer Predicti...mentioning

confidence: 99%

The Current and Evolving Landscape of Breast Cancer Prediction and Prognosis

Sharma

2023

MRAJ

View full text Add to dashboard Cite

The biological behaviour of breast cancer is remarkably heterogeneous and it is essential to have tools which can provide the necessary risk stratification to plan clinical management. Breast cancer prediction and prognosis needs to be holistic, and account for multiple levels of organisation. The histological classification and grading of the tumour itself presents valuable predictive and prognostic information. Hormone receptor status remains a mainstay, but roles may emerge for assessment of the intrinsic molecular subtype, for a molecular subclassification of triple negative carcinomas, and for whole genome sequencing. The recent discovery that antibody drug conjugates are effective in patients with weak HER-2 protein expression has led to the definition of the HER-2 low group. There has been a proliferation in predictive and prognostic models, numbering over 900, but the majority are at high risk of bias and tend to perform less well when applied to populations beyond the development cohort. The Nottingham Prognostic Index is a notable exception. Of the molecular risk stratification tools currently available, Oncotype Dx is the most widely recommended and used, but the question as to which test is superior remains unanswerable with current data. There is growing interest in omics-based approaches from which a number of biomarkers are being developed. It is well established that the microenvironment of the tumour is key to the tumour’s behaviour. Some components contain and destroy the cancer, whereas others are co-opted by the tumour and aid in its progression; the current evidence is reviewed, including the current status of tumour infiltrating lymphocyte assessment and immune checkpoint inhibition in breast cancer. The use of the liquid biopsy to achieve early detection of tumours and to manage tumour evolution is receiving intense attention; approaches include circulating tumour cells and circulating tumour DNA. Specific assessment of tumour giant cells may also provide the ability to anticipate tumour evolution. The influence of the gut microbiome on breast cancer is an intriguing development which requires further intensive study. There is a paucity of biomarkers in the setting of hereditary breast cancer. The use of polygenic risk scores in this setting is an interesting development requiring further study. The greatest challenge of all is to pull from such complexity key decision nodes that are clear enough to guide treatment decisions without losing the depth and richness of the information that underlies them. Seeking and finding this balance has been and will continue to be the holy grail of all endeavours in this field.

show abstract

“…However, traditional risk factor-based models, relying on methods like logistic regression or Cox regression, have low discrimination accuracy with the area under the receiver operating characteristic curve (AUC) between 0.53 and 0.64. 7 Some models heavily rely on family history and lack generalizability, and others can be biased when applied to specific subpopulations 8,9 Moreover, nontraditional factors, such as chronic diseases, are not usually included in the models, although chronic conditions are believed to raise cancer risk as much as lifestyle does 10 . Therefore, new methods and models are urgently needed to improve cancer risk predictions and facilitate the development of effective cancer prevention strategies.…”

Section: Introductionmentioning

confidence: 99%

Analyzing the Performance of Explainable Machine Learning Models in Risk Factor Identification for Major Cancers (Preprint)

Huang

2024

Preprint

View full text Add to dashboard Cite

BACKGROUND Cancer is a life-threatening disease and a leading cause of death worldwide, with an estimated 611,000 deaths and over 2 million new cases in the United States in 2024. The rising incidence of major cancers, including among younger individuals, highlights the need for early screening and monitoring of risk factors to manage and decrease cancer risk. OBJECTIVE To identify pivotal factors essential for predicting the risk factors for four major cancer types (breast, colorectal, lung, and prostate) through the utilization of explainable machine learning techniques is imperative due to the increasing burden of cancer patients. METHODS De-identified electronic health record data from MIMIC-III was used to identify patients with four types of cancer who had longitudinal hospital visits prior to receiving a cancer diagnosis. Their records were matched and combined with those of patients without cancer diagnoses using propensity scores based on demographic factors. Three advanced models, penalized Logistic Regression (LR), Random Forest (RF), and Multilayer Perceptron (MLP), were conducted to identify the rank of risk factors for each cancer type, with feature importance analysis for RF and MLP models. The Rank Biased Overlap was adopted to compare the similarity of ranked risk factors across cancer types. RESULTS Our framework evaluated the prediction performance of explainable ML models, in which MLP achieved an AUC of 0.78 for breast cancer, 0.76 for colorectal cancer, 0.84 for lung cancer, and 0.78 for prostate cancer, respectively. In addition to demographic risk factors, the most prominent non-traditional risk factors overlapped across models and cancer types, including hyperlipidemia, diabetes, depressive disorders, heart diseases, and anemia. The similarity analysis indicated the unique risk factor pattern for lung cancer from other cancer types. CONCLUSIONS The study's findings demonstrate the effectiveness of explainable ML models in predicting non-traditional risk factors for major cancers and highlight the importance of considering unique risk profiles for different cancer types. These insights may contribute to efficient cancer screening and tailored cancer prevention strategies, which, in turn, offer fundamental support for clinical decision-making processes.

show abstract

An Assessment of the Predictive Performance of Current Machine Learning–Based Breast Cancer Risk Prediction Models: Systematic Review

Cited by 7 publications

References 46 publications

Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma

Artificial Intelligence in the Diagnosis and Treatment of Pancreatic Cystic Lesions and Adenocarcinoma

The Current and Evolving Landscape of Breast Cancer Prediction and Prognosis

Analyzing the Performance of Explainable Machine Learning Models in Risk Factor Identification for Major Cancers (Preprint)

Contact Info

Product

Resources

About