Preoperative weight loss interventions before total hip and knee arthroplasty: a systematic review of randomized controlled trials.

BACKGROUND: The high co-prevalence of obesity and end-stage osteoarthritis requiring arthroplasty, with the former being a risk factor for complications during arthroplasty, has led to increasing interest in employing preoperative weight loss interventions such as bariatric surgery and diet modification. However, the current evidence is conflicting, and this study aimed to investigate the effect of weight loss intervention before arthroplasty in prospective randomized controlled trials. METHODS: Four electronic databases (MEDLINE, EMBASE, Web of Science, and Cochrane Central Register of Controlled Trials) were searched for prospective randomized controlled trials that compared weight loss interventions with usual care from inception to October 2023 by following the PRISMA guidelines. The Cochrane risk of bias tool and GRADE framework were used to assess the quality of the studies. Meta-analyses were performed when sufficient data were available from 2 or more studies. RESULTS: Three randomized controlled trials involving 198 patients were identified. Two studies employed diet modification, and one study utilized bariatric surgery. All three studies reported significant reductions in body weight and body mass index (BMI), and intervention groups had fewer postoperative complications. There was no difference in the length of stay between the intervention group and the control group. Variable patient-reported outcome measures were used by different research groups. CONCLUSION: Weight loss intervention can achieve significant reductions in body weight and body mass index before arthroplasty, with fewer postoperative complications reported. Further studies with different populations could confirm the effect of these interventions among populations with different obesity characteristics.

Cancer-Causing Effects of Orthopaedic Metal Implants in Total Hip Arthroplasty.

BACKGROUND: Metal implants have been preferentially used in THA due to its biocompatibility, mechanical stability and durability. Yet concerns have emerged regarding their potential to release metallic ions, leading to long-term adverse effects, including carcinogenicity. This study aimed to investigate the risk of cancer development in patients with orthopaedic metal implants in total hip arthroplasty (THA). METHODS: Patients with THA conducted at a local tertiary implant centre from 2001-2008 were linked to the local cancer registry and followed up to the end of 2023. Standardized incidence ratios (SIRs) for cancer incidence and its confidence interval by Poisson distribution were calculated. Survival analysis was depicted using the Kaplan-Meier method, and the log-rank test was used to assess the differences across groups. RESULTS: The study cohort included 388 patients and 53 cancers diagnosed during follow-up, at least 5 years post THA. All-site cancer risks were increased in patients with THA (SIR: 1.97; 95% CI: 1.48-2.46), validated with chi-square analysis (chi-square = 15.2551, N = 100,388, p < 0.01). A statistically significant increase in multiple site-specific cancers including haematological cancers were identified. CONCLUSIONS: Patients with THA were found to have an increased risk for cancer compared to the general population during a mean follow-up of 16 years.

Circulating Tumour Cells in the Prediction of Bone Metastasis.

Bone is the most common organ for the development of metastases in many primary tumours, including those of the breast, prostate and lung. In most cases, bone metastasis is incurable, and treatment is predominantly palliative. Much research has focused on the role of Circulating Tumour Cells (CTCs) in the mechanism of metastasis to the bone, and methods have been developed to isolate and count CTCs from peripheral blood. Several methods are currently being used in the study of CTCs, but only one, the CellSearchTM system has been approved by the United States Food and Drug Administration for clinical use. This review summarises the advantages and disadvantages, and outlines which clinical studies have used these methods. Studies have found that CTC numbers are predictive of bone metastasis in breast, prostate and lung cancer. Further work is required to incorporate information on CTCs into current staging systems to guide treatment in the prevention of tumour progression into bone.

Bayesian Disease Mapping to Identify High-Risk Population for Oral Cancer: A Retrospective Spatiotemporal Analysis.

Objectives: Bayesian mapping is an effective spatiotemporal approach to identify high-risk geographic areas for diseases and has not been used to identify oral cancer hotspots in Australia previously. This retrospective disease mapping study was undertaken to identify the oral cancer trends and patterns within the Queensland state in Australia. Methods: This study included data obtained from Queensland state Cancer Registry from 1982 to 2018. Domains mapped included the oral cancer incidence and mortality in Queensland (QLD). Local government areas (LGAs) and suburbs were utilized as geographical units for the estimation using Bayesian mapping approach. Results: Of the 78 LGAs, 21 showed high-oral cancer incidence as measured using higher median smoothed incidence risk (SIR), above the state average. Specifically, nine LGAs within predominantly rural areas had SIR above 100% of the state average. Of these, only one LGA (Mount Isa City) had a median smoothed SIR and 95% CI of 2.61 (2.14-3.15) which was constantly above 100% of the state average. Furthermore, mortality risk estimated using smoothed mortality risk (SMR), were significantly higher than the state average in 31 LGAs. Seventeen LGAs had a median SMR above 100% of the state average while three LGAs had the highest overall, 3- and 5-year mortality risks. Considering the 95% credible interval which is indicative of the uncertainty around the estimates, three LGAs had the highest overall mortality risks-Yarrabah Aboriginal Shire (3.80 (2.16-6.39)), Cook Shire (3.37 (2.21-5.06)), and Mount Isa City (3.04 (2.40-3.80)). Conclusion: Bayesian disease mapping approach identified multiple incidence and mortality hotspots within regional areas of the Queensland. Findings from our study can aid in designing targeted public health screening and interventions for primary prevention of oral cancer in regional and remote communities.

Oral cavity and oropharyngeal carcinoma disparities in age and survival in Indigenous and non-Indigenous populations of Queensland.

OBJECTIVES: To investigate the risk and prognosis of oral squamous cell carcinoma (SCC) between Indigenous and non-Indigenous populations of Queensland. MATERIALS AND METHODS: Retrospective analysis of data from the Queensland Cancer Registry (QCR) between the years 1982-2018. Main outcome measures were age at diagnosis and cumulative survival to compare the risk and prognosis of oral SCC between the populations. RESULTS: 9424 patients with self-declared ethnicity were identified with oral SCC from the QCR, with a male to female ratio of 2.56:1. Of these patients, 9132 were non-Indigenous (96.9%) and 292 Indigenous (3.1%). Indigenous people were significantly younger at diagnosis (mean (SD) age 54.3 (10.1) years), compared to 62.0 (12.1) years in non-Indigenous people. Mean survival in the full cohort was 4.3 years (SD: 5.6), with Indigenous people presenting a significant shorter mean survival of 2.0 years (SD: 3.5) when compared with 4.4 years (SD: 5.7) in non-Indigenous people (p < 0.001). CONCLUSIONS: Indigenous Australians are diagnosed at a significantly younger age and present with worse survival and poorer prognosis. Due to missing variables in the Queensland Cancer Registry, it is not possible in the current study to ascertain the scientific or social reasons behind these disparities. CLINICAL RELEVANCE: Results from this study can inform public policy and raise awareness in Queensland regarding disparity in oral cancer prognosis.

Predicting oral cancer survival-Development and validation of an Asia-Pacific nomogram.

BACKGROUND: Nomograms are graphical calculating devices that predict response to treatment during cancer management. Oral squamous cell carcinoma (OSCC) is a lethal and deforming disease of rising incidence and global significance. The aim of this study was to develop a nomogram to predict individualized OSCC survival using a population-based dataset obtained from Queensland, Australia and externally validated using a cohort of OSCC patients treated in Hong Kong. METHODS: Clinico-pathological data for newly diagnosed OSCC patients, including age, sex, tumour site and grading, were accessed retrospectively from the Queensland Cancer Registry (QCR) in Australia and the Clinical Data Analysis and Reporting System (CDARS) in Hong Kong. Multivariate Cox proportional hazard regression was used to construct overall survival (OS) and cancer-specific survival (CSS) prediction models. Nomograms were internally validated using 10-fold cross validation, and externally validated against the Hong Kong dataset. RESULTS: Data from 9885 OSCC patients in Queensland and 465 patients from Hong Kong were analysed. All clinico-pathological variables significantly influenced survival outcomes. Nomogram calibration curves demonstrated excellent agreement between predicted and actual probability for Queensland patients. External validation in the Hong Kong population demonstrated slightly poorer nomogram performance, but predictive power remained strong. CONCLUSION: Based upon readily available data documenting patient demographic and clinico-pathological variables, predictive nomograms offer pragmatic aid to clinicians in individualized treatment planning and prognosis assessment in contemporary OSCC management.

Influence of time between surgery and adjuvant radiotherapy on prognosis for patients with head and neck squamous cell carcinoma: A systematic review.

The timing of postoperative radiotherapy following surgical intervention in patients with head and neck cancer remains a controversial issue. This review aims to summarize findings from available studies to investigate the influence of time delays between surgery and postoperative radiotherapy on clinical outcomes. Articles between 1 January 1995 and 1 February 2022 were sourced from PubMed, Web of Science, and ScienceDirect. Twenty-three articles met the study criteria and were included; ten studies showed that delaying postoperative radiotherapy might negatively impact patients and lead to a poorer prognosis. Delaying the start time of radiotherapy, 4 weeks after surgery did not result in poorer prognoses for patients with head and neck cancer, although delays beyond 6 weeks might worsen patients' overall survival, recurrence-free survival, and locoregional control. Prioritization of treatment plans to optimize the timing of postoperative radiotherapy regimes is recommended.

Oral cancer in Australia: Rising incidence and worsening mortality.

BACKGROUND: Oral cancer, predominantly squamous cell carcinoma (SCC), is a lethal and deforming disease of rising incidence. Although largely preventable by eliminating harmful tobacco and alcohol risk factor behaviour, 5-year survival rates remain around 50%, primarily due to late presentation of advanced stage disease. Whilst low socio-economic status, regional and remote location and indigenous status are associated with head and neck cancer in general, detailed incidence and demographic data for oral SCC in Australia are limited. This study aimed to characterise the Queensland population at risk of oral SCC development. METHODS: Following ethical approval, the Queensland Cancer Register (QCR) dataset was analysed to determine patterns of incidence, anonymised patient demographics, clinical presentation and outcome data for oral SCC cases diagnosed between 1982 and 2018. RESULTS: Data from 9887 patients were obtained. Mean age at diagnosis was 64.55 years, with a male-to-female ratio of 2.51:1; males were diagnosed at a younger age (p < 0.001). At study census date, 59% of patients had died, with females demonstrating longer mean survival (p < 0.001). Clinicopathological data confirmed that SCC most commonly arose from tongue sites (49%) and, whilst tumours were predominantly moderately differentiated in nature (63%), patients with poorly differentiated carcinomas exhibited shortest survival times (p < 0.05). Over the 36-year study period, the number of diagnoses increased 4.49-fold, whilst the number of deaths increased 19.14-fold. CONCLUSION: Oral SCC poses a significant and growing healthcare problem in Queensland. In the absence of national screening, characterising the high-risk oral SCC population facilitates pragmatic opportunities to raise disease awareness, to deliver targeted screening and effective primary prevention strategies, and to provide early interventional treatment intervention to reduce disease mortality and morbidity.

The Impact of Surgery on Circulating Malignant Tumour Cells in Oral Squamous Cell Carcinoma.

IMPORTANCE: The extent to which surgical management of oral squamous cell carcinoma (OSCC) disseminates cancer is currently unknown. OBJECTIVE: To determine changes in numbers of malignant cells released into systemic circulation immediately following tumour removal and over the first seven post-operative days. DESIGN: An observational study from March 2019 to February 2021. SETTING: This study was undertaken at Queen Mary University Hospital, Hong Kong. PARTICIPANTS: Patients with biopsy-proven oral SCC were considered for eligibility. Patients under 18 years of age, pregnant or lactating women and those unable to understand the study details or unable to sign the consent form were excluded. Twenty-two patients were enrolled (12 male and 10 female) with mean age of 65.5 years. INTERVENTION: Primary tumour management was performed in accord with multi-disciplinary team agreement. Anaesthesia and post-operative care were unaltered and provided in accord with accepted clinical practice. MAIN OUTCOMES AND MEASURES: Three types of malignant cells detected in peripheral blood samples were enumerated and sub-typed based on the presence of chromosomal aneuploidy and immunohistochemical characteristics. To test the hypothesis that malignant cells are released by surgery, the numbers of single circulating tumour cells (CTCs), circulating tumour microemboli (CTM) and circulating endothelial cells (CTECs) were recorded pre-operatively, upon tumour removal and the second and seventh post-operative days. RESULTS: Of a potential 88 data collection points, specimens were not obtainable in 12 instances. Tumour removal resulted in a statistically significant increase in CTCs and a non-statistically significant rise in CTMs. CTCs, CTMs and CTECs were detected in the majority of patients up to the seventh post-operative day. Individual patients demonstrated striking increases in post-operative CTCs and CTECs numbers. CONCLUSIONS/RELEVANCE: Surgical management of OSCC has a significant impact on the systemic distribution of cancer cells. Malignant cells persisted post-operatively in a manner independent of recognised staging methods suggesting differences in tumour biology between individuals. Further investigation is warranted to determine whether circulating malignant cell enumeration can be used to refine risk stratification for patients with OSCC.

Explainable ensemble learning model improves identification of candidates for oral cancer screening.

OBJECTIVES: Artificial intelligence could enhance the use of disparate risk factors (crude method) for better stratification of patients to be screened for oral cancer. This study aims to construct a meta-classifier that considers diverse risk factors to identify patients at risk of oral cancer and other suspicious oral diseases for targeted screening. MATERIALS AND METHODS: A retrospective dataset from a community oral cancer screening program was used to construct and train the novel voting meta-classifier. Comprehensive risk factor information from this dataset was used as input features for eleven supervised learning algorithms which served as base learners and provided predicted probabilities that are weighted and aggregated by the meta-classifier. Training dataset was augmented using SMOTE-ENN. Additionally, Shapley additive explanations (SHAP) values were generated to implement the explainability of the model and display the important risk factors. RESULTS: Our meta-classifier had an internal validation recall, specificity, and AUROC of 0.83, 0.86, and 0.85 for identifying the risk of oral cancer and 0.92, 0.60, and 0.76 for identifying suspicious oral mucosal disease respectively. Upon external validation, the meta-classifier had a significantly higher AUROC than the crude/current method used for identifying the risk of oral cancer (0.78 vs 0.46; p = 0.001) Also, the meta-classifier had better recall than the crude method for predicting the risk of suspicious oral mucosal diseases (0.78 vs 0.47). CONCLUSION: Overall, these findings showcase that our approach optimizes the use of risk factors in identifying patients for oral screening which suggests potential clinical application.

Predicting Overall Survival Using Machine Learning Algorithms in Oral Cavity Squamous Cell Carcinoma.

BACKGROUND/AIM: Machine learning (ML) models are often modelled to predict cancer prognosis but rarely consider spatial factors in a region. Hence this study explored machine learning algorithms utilising Local Government Areas (LGAs) in Queensland, Australia to spatially predict 3- and 5-year prognosis of oral cancer patients and provide clinical interpretability of the predicted outcome made by the ML model. PATIENTS AND METHODS: Data from a total of 3,841 oral cancer patients were retrieved from the Queensland Cancer Registry (QCR). Synthesizing minority oversampling technique together with edited nearest neighbours (SMOTE-ENN) was used to pre-process unbalanced datasets. Five ML models: logistic regression, random forest classifier, XGBoost, Gaussian Naïve Bayes and Voting Classifier were trained. Predictive features were age, sex, LGAs, tumour site and differentiation. Outcomes were 3- and 5-year overall survival of patients. Model performances on test set were evaluated using area under the curve and F1 scores. SHapley Additive exPlanations (SHAP) method was applied to the best performing model for model interpretation of the predicted outcome. RESULTS: The Voting Classifier was the best performing model with F1 score of 0.58 and 0.64 for 3- and 5-year overall survival, respectively. Age was the most important feature in the Voting Classifier in 3- and 5-year prognosis prediction. LGAs at diagnosis was the top 3 predictive feature for both 3- and 5-year models. CONCLUSION: The Voting Classifier demonstrated the best overall performance in classifying both 3- and 5-year overall survival of oral cancer patients in Queensland. SHAP method provided clinical understanding of the predictive features of the Voting Classifier.

Machine Learning-Based Genome-Wide Salivary DNA Methylation Analysis for Identification of Noninvasive Biomarkers in Oral Cancer Diagnosis.

This study aims to examine the feasibility of ML-assisted salivary-liquid-biopsy platforms using genome-wide methylation analysis at the base-pair and regional resolution for delineating oral squamous cell carcinoma (OSCC) and oral potentially malignant disorders (OPMDs). A nested cohort of patients with OSCC and OPMDs was randomly selected from among patients with oral mucosal diseases. Saliva samples were collected, and DNA extracted from cell pellets was processed for reduced-representation bisulfite sequencing. Reads with a minimum of 10× coverage were used to identify differentially methylated CpG sites (DMCs) and 100 bp regions (DMRs). The performance of eight ML models and three feature-selection methods (ANOVA, MRMR, and LASSO) were then compared to determine the optimal biomarker models based on DMCs and DMRs. A total of 1745 DMCs and 105 DMRs were identified for detecting OSCC. The proportion of hypomethylated and hypermethylated DMCs was similar (51% vs. 49%), while most DMRs were hypermethylated (62.9%). Furthermore, more DMRs than DMCs were annotated to promoter regions (36% vs. 16%) and more DMCs than DMRs were annotated to intergenic regions (50% vs. 36%). Of all the ML models compared, the linear SVM model based on 11 optimal DMRs selected by LASSO had a perfect AUC, recall, specificity, and calibration (1.00) for OSCC detection. Overall, genome-wide DNA methylation techniques can be applied directly to saliva samples for biomarker discovery and ML-based platforms may be useful in stratifying OSCC during disease screening and monitoring.

Oral Squamous Cell Carcinoma Frequency in Young Patients from Referral Centers Around the World.

Oral squamous cell carcinoma (OSCC) commonly affects older patients; however, several studies have documented an increase in its incidence among younger patients. Therefore, it is important to investigate if this trend is also found in different geographic regions. The pathology files of diagnostic and therapeutic institutions from different parts of the globe were searched for OSCC cases diagnosed from 1998 to 2018. Data regarding the sex, age, and tumor location of all cases, as well as the histologic grade and history of exposure to risk habits of cases diagnosed as OSCC in young patients (≤ 40 years of age) were obtained. The Chi-square test was used to determine any increasing trend. A total of 10,727 OSCC cases were identified, of which 626 cases affected young patients (5.8%). Manipal institution (India) showed the highest number of young patients (13.2%). Males were the most affected in both age groups, with the tongue and floor of the mouth being the most affected subsites. OSCC in young individuals were usually graded as well or moderately differentiated. Only 0.9% of the cases occurred in young patients without a reported risk habit. There was no increasing trend in the institutions and the period investigated (p > 0.05), but a decreasing trend was observed in Hong Kong and the sample as a whole (p < 0.001). In conclusion there was no increase of OSCC in young patients in the institutions investigated and young white females not exposed to any known risk factor represented a rare group of patients affected by OSCC.

Performance of a simplified scoring system for risk stratification in oral cancer and oral potentially malignant disorders screening.

BACKGROUND: Impact and efficiency of oral cancer and oral potentially malignant disorders screening are most realized in "at-risk" individuals. However, tools that can provide essential knowledge on individuals' risks are not applied in risk-based screening. This study aims to optimize a simplified risk scoring system for risk stratification in organized oral cancer and oral potentially malignant disorders screening. METHODS: Participants were invited to attend a community-based oral cancer and oral potentially malignant disorders screening program in Hong Kong. Visual oral examination was performed for all attendees and information on sociodemographic characteristics as well as habitual, lifestyle, familial, and comorbidity risk factors were obtained. Individuals' status of those found to have suspicious lesions following biopsy and histopathology were classified as positive/negative and this outcome was used in a multiple logistic regression analysis with variables collected during screening. Odds ratio weightings were then used to develop a simplified risk scoring system which was validated in an external cohort. RESULTS: Of 979 participants, 4.5% had positive status following confirmatory diagnosis. A 12-variable simplified risk scoring system with weightings was generated with an AUC, sensitivity, and specificity of 0.82, 0.71, and 0.78 for delineating high-risk cases. Further optimization on the validation cohort of 491 participants yielded a sensitivity and specificity of 0.75 and 0.87 respectively. CONCLUSIONS: The simplified risk scoring system was able to stratify oral cancer and oral potentially malignant disorders risk with satisfactory sensitivity and specificity and can be applied in risk-based disease screening.

Oral Cancer Awareness and Individuals' Inclination to Its Screening and Risk Prediction in Hong Kong.

Assessing the baseline knowledge status and expectations of the target population of any health promotion and secondary prevention program is essential to the success of such intervention. To obtain this information about the Hong Kong population a priori to implementing these preventive strategies for oral cancer in addition to determining the willingness of potential screening participants to take risk-profiling assessments, a cross-sectional survey was conducted between November 2019 and March 2020. A total of 964 residents between the ages 18 and 86 years were invited to participate in this study across the three geographical areas in Hong Kong. Most participants self-reported being aware of oral cancer (86.3%), although the proportion of those with substantial knowledge on salient risk factors and early identifiable signs were very low (2.9%). Age and level of education were the only demographic characteristics associated with the knowledge status. The proportion of participants willing to attend community screening and partake in risk profiling assessment was high (83.9% and 80.9% respectively). Willingness to attend community screening was directly associated with respondents' self-reported oral cancer awareness status (OR: 1.9, 95% CI: 1.22-2.96). Also, we observed that those participants who were willing to attend screening are more inclined to take risk prediction assessments that those not willing to attend. These findings have showcased the need to intensify health promotion via personal skills development to encourage early disease presentation and will assist in the planning of these programs accordingly in the Hong Kong population.

Efficacy of hypermethylated DNA biomarkers in saliva and oral swabs for oral cancer diagnosis: Systematic review and meta-analysis.

OBJECTIVES: This study aims to determine the diagnostic test accuracy (DTA) of hypermethylated DNA biomarkers in saliva and oral swabs for oral squamous cell carcinoma (OSCC) detection from the prevalidation studies available. MATERIALS AND METHODS: Electronic database searching of PubMed, EMBASE, Cochrane Library, Scopus, Web of Science, and LILACS was conducted to identify relevant articles that were published between January 1, 2000, and August 1, 2020. RESULTS: Meta-analysis was conducted based on 11 of 20 studies selected for review. Included studies had high bias concerns on the QUADAS-2 study assessment tool. We found that salivary and oral swab hypermethylation markers had better specificity than sensitivity for oral cancer detection. Summary sensitivity and specificity (95% CI) of hypermethylation panels were 86.2% (60-96.2) and 90.6% (85.9-93.9) while for individual markers, summary sensitivity and specificity (95% CI) were 70% (56.9-80.5) and 91.9% (80.3-96.9), respectively. Respective positive and negative likelihood ratios for combined markers were 9.2 (5.89-14.36) and 0.15 (0.05-0.5), and 8.61 (3.39-21.87) and 0.33 (0.22-0.49) for single-application biomarkers. CONCLUSION: DNA hypermethylation biomarkers especially in combination have acceptable DTA that warrants further optimization with rigorous biomarker evaluation methods for conclusive determination of their efficacy.

A comparison of two methods for the detection of circulating tumour cells in patients with oral cavity cancer.

BACKGROUND: Circulating tumour cells (CTCs) detected in patient blood samples are relevant as diagnostic and prognostic markers offering insights into tumour behaviour and guiding treatment of cancer at an individualised level. The aim of this study was to ascertain the feasibility of detecting CTCs in oral squamous cell carcinoma (OSCC) using two different methods so as to determine the optimal method for the study of this cancer. METHODS: Comparison of the numbers of CTCs, circulating tumour micro-emboli (CTMs) and circulating tumour endothelial cells (CTECs), was undertaken in forty clinical samples of oral squamous cell carcinoma (OSCC) determined by filtration (ISET® ) and in situ fluorescent immunostaining (i-FISH, Cytelligen® ) immunostaining and in situ hybridisation. RESULTS: i-FISH detected CTCs in 80% of samples compared with 40% of samples analysed by microfiltration. i-FISH detected CTCs in a further 40% of samples in which microfiltration did not detect CTCs. No CTC clusters were detected by microfiltration while i-FISH detected CTM in 12.5% of samples. i-FISH analysis detected CTECs in 20/40 samples. CONCLUSION: These results highlight significant differences in detection of CTCs, CTM and CTECs between i-FISH and microfiltration when applied to OSCC samples, suggesting that technologies capable of detecting circulating aneuploid cells more accurately detect CTCs. i-FISH also detected CTM and CTEC not detected using ISET® . With proven prognostic relevance in adenocarcinomas, accurate enumeration of CTCs, CTMs and CTECs may be a clinically useful tool in the management of OSCC and may aid in the reduction of false-negative diagnoses.

Comparison of time-to-event machine learning models in predicting oral cavity cancer prognosis.

BACKGROUND: Applying machine learning to predicting oral cavity cancer prognosis is important in selecting candidates for aggressive treatment following diagnosis. However, models proposed so far have only considered cancer survival as discrete rather than dynamic outcomes. OBJECTIVES: To compare the model performance of different machine learning-based algorithms that incorporate time-to-event data. These algorithms included DeepSurv, DeepHit, neural net-extended time-dependent cox model (Cox-Time), and random survival forest (RSF). MATERIALS AND METHODS: Retrospective cohort of 313 oral cavity cancer patients were obtained from electronic health records. Models were trained on patient data following preprocessing. Predictors were based on demographic, clinicopathologic, and treatment information of the cases. Outcomes were the disease-specific and overall survival. Multivariable analyses were conducted to select significant prognostic features associated with tumor prognosis. Two models were generated per algorithm based on all-prognostic features and significant-prognostic features following statistical analysis. Concordance index (c-index) and integrated Brier scores were used as performance evaluators and model stability was assessed using intraclass correlation coefficients (ICC) calculated from these measures obtained from the cross-validation folds. RESULTS: While all models were satisfactory, better discriminatory performance and calibration was observed for disease-specific than overall survival (mean c-index: 0.85 vs 0.74; mean integrated Brier score: 0.12 vs 0.17). DeepSurv performed best in terms of discrimination for both outcomes (c-indices: 0.76 -0.89) while RSF produced better calibrated survival estimates (integrated Brier score: 0.06 -0.09). Model stability of the algorithms varied with the outcomes as Cox-Time had the best intraclass correlation coefficient (mean ICC: 1.00) for disease-specific survival while DeepSurv was most stable for overall survival prediction (mean ICC: 0.99). CONCLUSIONS: Machine learning algorithms based on time-to-event outcomes are successful in predicting oral cavity cancer prognosis with DeepSurv and RSF producing the best discriminative performance and calibration.

Deep Learning Predicts the Malignant-Transformation-Free Survival of Oral Potentially Malignant Disorders.

Machine-intelligence platforms for the prediction of the probability of malignant transformation of oral potentially malignant disorders are required as adjunctive decision-making platforms in contemporary clinical practice. This study utilized time-to-event learning models to predict malignant transformation in oral leukoplakia and oral lichenoid lesions. A total of 1098 patients with oral white lesions from two institutions were included in this study. In all, 26 features available from electronic health records were used to train four learning algorithms-Cox-Time, DeepHit, DeepSurv, random survival forest (RSF)-and one standard statistical method-Cox proportional hazards model. Discriminatory performance, calibration of survival estimates, and model stability were assessed using a concordance index (c-index), integrated Brier score (IBS), and standard deviation of the averaged c-index and IBS following training cross-validation. This study found that DeepSurv (c-index: 0.95, IBS: 0.04) and RSF (c-index: 0.91, IBS: 0.03) were the two outperforming models based on discrimination and calibration following internal validation. However, DeepSurv was more stable than RSF upon cross-validation. External validation confirmed the utility of DeepSurv for discrimination (c-index-0.82 vs. 0.73) and RSF for individual survival estimates (0.18 vs. 0.03). We deployed the DeepSurv model to encourage incipient application in clinical practice. Overall, time-to-event models are successful in predicting the malignant transformation of oral leukoplakia and oral lichenoid lesions.