FACE-Q for Measuring Patient-reported Outcomes after Facial Skin Cancer Surgery: Cross-cultural Validation.

Background: Facial skin cancer and its surgical treatment can affect health-related quality of life. The FACE-Q Skin Cancer Module is a patient-reported outcome measure that measures different aspects of health-related quality of life and has recently been translated into Dutch. This study aimed to evaluate the performance of the translated version in a Dutch cohort using modern psychometric measurement theory (Rasch). Methods: Dutch participants with facial skin cancer were prospectively recruited and asked to complete the translated FACE-Q Skin Cancer Module. The following assumptions of the Rasch model were tested: unidimensionality, local independence, and monotonicity. Response thresholds, fit statistics, internal consistency, floor and ceiling effects, and targeting were assessed for all scales and items within the scales. Responsiveness was tested for the "cancer worry" scale. Results: In total, 259 patients completed the preoperative questionnaire and were included in the analysis. All five scales assessed showed a good or sufficient fit to the Rasch model. Unidimensionality and monotonicity were present for all scales. Some items showed a local dependency. Most of the scales demonstrate ordered item thresholds and appropriate fit statistics. Conclusions: The FACE-Q Skin Cancer Module is a well-designed patient-reported outcome measure that shows psychometric validity for the translated version in a Dutch cohort, using classical and modern test theory.

Ultrasound Radiomics Features to Identify Patients With Triple-Negative Breast Cancer: A Retrospective, Single-Center Study.

OBJECTIVES: Patients with triple-negative breast cancer (TNBC) exhibit a fast tumor growth rate and poor survival outcomes. In this study, we aimed to develop and compare intelligent algorithms using ultrasound radiomics features in addition to clinical variables to identify patients with TNBC prior to histopathologic diagnosis. METHODS: We used single-center, retrospective data of patients who underwent ultrasound before histopathologic verification and subsequent neoadjuvant systemic treatment (NAST). We developed a logistic regression with an elastic net penalty algorithm using pretreatment ultrasound radiomics features in addition to patient and tumor variables to identify patients with TNBC. Findings were compared to the histopathologic evaluation of the biopsy specimen. The main outcome measure was the area under the curve (AUC). RESULTS: We included 1161 patients, 813 in the development set and 348 in the validation set. Median age was 50.1 years and 24.4% (283 of 1161) had TNBC. The integrative model using radiomics and clinical information showed significantly better performance in identifying TNBC compared to the radiomics model (AUC: 0.71, 95% confidence interval [CI]: 0.65-0.76 versus 0.64, 95% CI: 0.57-0.71, P = .004). The five most important variables were cN status, shape surface volume ratio (SA:V), gray level co-occurrence matrix (GLCM) correlation, gray level dependence matrix (GLDM) dependence nonuniformity normalized, and age. Patients with TNBC were more often categorized as BI-RADS 4 than BI-RADS 5 compared to non-TNBC patients (P = .002). CONCLUSION: A machine learning algorithm showed promising potential to identify patients with TNBC using ultrasound radiomics features and clinical information prior to histopathologic evaluation.

Minimally Invasive Breast Biopsy After Neoadjuvant Systemic Treatment to Identify Breast Cancer Patients with Residual Disease for Extended Neoadjuvant Treatment: A New Concept.

BACKGROUND: Breast cancer patients with residual disease after neoadjuvant systemic treatment (NAST) have a worse prognosis compared with those achieving a pathologic complete response (pCR). Earlier identification of these patients might allow timely, extended neoadjuvant treatment strategies. We explored the feasibility of a vacuum-assisted biopsy (VAB) after NAST to identify patients with residual disease (ypT+ or ypN+) prior to surgery. METHODS: We used data from a multicenter trial, collected at 21 study sites (NCT02948764). The trial included women with cT1-3, cN0/+ breast cancer undergoing routine post-neoadjuvant imaging (ultrasound, MRI, mammography) and VAB prior to surgery. We compared the findings of VAB and routine imaging with the histopathologic evaluation of the surgical specimen. RESULTS: Of 398 patients, 34 patients with missing ypN status and 127 patients with luminal tumors were excluded. Among the remaining 237 patients, tumor cells in the VAB indicated a surgical non-pCR in all patients (73/73, positive predictive value [PPV] 100%), whereas PPV of routine imaging after NAST was 56.0% (75/134). Sensitivity of the VAB was 72.3% (73/101), and 74.3% for sensitivity of imaging (75/101). CONCLUSION: Residual cancer found in a VAB specimen after NAST always corresponds to non-pCR. Residual cancer assumed on routine imaging after NAST corresponds to actual residual cancer in about half of patients. Response assessment by VAB is not safe for the exclusion of residual cancer. Response assessment by biopsies after NAST may allow studying the new concept of extended neoadjuvant treatment for patients with residual disease in future trials.

Can multi-modal radiomics using pretreatment ultrasound and tomosynthesis predict response to neoadjuvant systemic treatment in breast cancer?

OBJECTIVES: Response assessment to neoadjuvant systemic treatment (NAST) to guide individualized treatment in breast cancer is a clinical research priority. We aimed to develop an intelligent algorithm using multi-modal pretreatment ultrasound and tomosynthesis radiomics features in addition to clinical variables to predict pathologic complete response (pCR) prior to the initiation of therapy. METHODS: We used retrospective data on patients who underwent ultrasound and tomosynthesis before starting NAST. We developed a support vector machine algorithm using pretreatment ultrasound and tomosynthesis radiomics features in addition to patient and tumor variables to predict pCR status (ypT0 and ypN0). Findings were compared to the histopathologic evaluation of the surgical specimen. The main outcome measures were area under the curve (AUC) and false-negative rate (FNR). RESULTS: We included 720 patients, 504 in the development set and 216 in the validation set. Median age was 51.6 years and 33.6% (242 of 720) achieved pCR. The addition of radiomics features significantly improved the performance of the algorithm (AUC 0.72 to 0.81; p = 0.007). The FNR of the multi-modal radiomics and clinical algorithm was 6.7% (10 of 150 with missed residual cancer). Surface/volume ratio at tomosynthesis and peritumoral entropy characteristics at ultrasound were the most relevant radiomics. Hormonal receptors and HER-2 status were the most important clinical predictors. CONCLUSION: A multi-modal machine learning algorithm with pretreatment clinical, ultrasound, and tomosynthesis radiomics features may aid in predicting residual cancer after NAST. Pending prospective validation, this may facilitate individually tailored NAST regimens. CLINICAL RELEVANCE STATEMENT: Multi-modal radiomics using pretreatment ultrasound and tomosynthesis showed significant improvement in assessing response to NAST compared to an algorithm using clinical variables only. Further prospective validation of our findings seems warranted to enable individualized predictions of NAST outcomes. KEY POINTS: • We proposed a multi-modal machine learning algorithm with pretreatment clinical, ultrasound, and tomosynthesis radiomics features to predict response to neoadjuvant breast cancer treatment. • Compared with the clinical algorithm, the AUC of this integrative algorithm is significantly higher. • Used prior to the initiative of therapy, our algorithm can identify patients who will experience pathologic complete response following neoadjuvant therapy with a high negative predictive value.

Development and Validation of a Novel Literature-Based Method to Identify Disparity-Sensitive Surgical Quality Metrics.

BACKGROUND: Disparity in surgical care impedes the delivery of uniformly high-quality care. Metrics that quantify disparity in care can help identify areas for needed intervention. A literature-based Disparity-Sensitive Score (DSS) system for surgical care was adapted by the Metrics for Equitable Access and Care in Surgery (MEASUR) group. The alignment between the MEASUR DSS and Delphi ratings of an expert advisory panel (EAP) regarding the disparity sensitivity of surgical quality metrics was assessed. STUDY DESIGN: Using DSS criteria MEASUR co-investigators scored 534 surgical metrics which were subsequently rated by the EAP. All scores were converted to a 9-point scale. Agreement between the new measurement technique (ie DSS) and an established subjective technique (ie importance and validity ratings) were assessed using the Bland-Altman method, adjusting for the linear relationship between the paired difference and the paired average. The limit of agreement (LOA) was set at 1.96 SD (95%). RESULTS: The percentage of DSS scores inside the LOA was 96.8% (LOA, 0.02 points) for the importance rating and 94.6% (LOA, 1.5 points) for the validity rating. In comparison, 94.4% of the 2 subjective EAP ratings were inside the LOA (0.7 points). CONCLUSIONS: Applying the MEASUR DSS criteria using available literature allowed for identification of disparity-sensitive surgical metrics. The results suggest that this literature-based method of selecting quality metrics may be comparable to more complex consensus-based Delphi methods. In fields with robust literature, literature-based composite scores may be used to select quality metrics rather than assembling consensus panels.

Enhanced Surgical Decision-Making Tools in Breast Cancer: Predicting 2-Year Postoperative Physical, Sexual, and Psychosocial Well-Being following Mastectomy and Breast Reconstruction (INSPiRED 004).

BACKGROUND: We sought to predict clinically meaningful changes in physical, sexual, and psychosocial well-being for women undergoing cancer-related mastectomy and breast reconstruction 2 years after surgery using machine learning (ML) algorithms trained on clinical and patient-reported outcomes data. PATIENTS AND METHODS: We used data from women undergoing mastectomy and reconstruction at 11 study sites in North America to develop three distinct ML models. We used data of ten sites to predict clinically meaningful improvement or worsening by comparing pre-surgical scores with 2 year follow-up data measured by validated Breast-Q domains. We employed ten-fold cross-validation to train and test the algorithms, and then externally validated them using the 11th site's data. We considered area-under-the-receiver-operating-characteristics-curve (AUC) as the primary metric to evaluate performance. RESULTS: Overall, between 1454 and 1538 patients completed 2 year follow-up with data for physical, sexual, and psychosocial well-being. In the hold-out validation set, our ML algorithms were able to predict clinically significant changes in physical well-being (chest and upper body) (worsened: AUC range 0.69-0.70; improved: AUC range 0.81-0.82), sexual well-being (worsened: AUC range 0.76-0.77; improved: AUC range 0.74-0.76), and psychosocial well-being (worsened: AUC range 0.64-0.66; improved: AUC range 0.66-0.66). Baseline patient-reported outcome (PRO) variables showed the largest influence on model predictions. CONCLUSIONS: Machine learning can predict long-term individual PROs of patients undergoing postmastectomy breast reconstruction with acceptable accuracy. This may better help patients and clinicians make informed decisions regarding expected long-term effect of treatment, facilitate patient-centered care, and ultimately improve postoperative health-related quality of life.

Physical well-being recovery trajectories by reconstruction modality in women undergoing mastectomy and breast reconstruction: Significant predictors and health-related quality of life outcomes.

OBJECTIVES: We sought to identify trajectories of patient-reported outcomes, specifically physical well-being of the chest (PWBC), in patients who underwent postmastectomy breast reconstruction, and further assessed its significant predictors, and its relationship with health-related quality of life (HRQOL). METHODS: We used data collected as part of the Mastectomy Reconstruction Outcomes Consortium study within a 2-year follow-up in 2012-2017, with 1422, 1218,1199, and 1417 repeated measures at assessment timepoints of 0,3,12, and 24 months, respectively. We performed latent class growth analysis (LCGA) in the implant group (IMPG) and autologous group (AUTOG) to identify longitudinal change trajectories, and then assessed its significant predictors, and its relationship with HRQOL by conducting multinomial logistic regression. RESULTS: Of the included 1424 patients, 843 were in IMPG, and 581 were in AUTOG. Both groups experienced reduced PWBC at follow-up. LCGA identified four distinct PWBC trajectories (χ2 = 1019.91, p<0.001): low vs medium high vs medium low vs high baseline PWBC that was restored vs. not-restored after 2 years. In 76.63%(n = 646) of patients in IMPG and 62.99% (n = 366) in AUTOG, PWBC was restored after two years. Patients in IMPG exhibited worse PWBC at 3 months post-surgery than that in AUTOG. Patients with low baseline PWBC that did not improve at 2-year follow up (n = 28, 4.82% for AUTOG) were characterized by radiation following reconstruction and non-white ethnicity. In IMPG, patients with medium low-restored trajectory were more likely to experience improved breast satisfaction, while patients developing high-restored trajectories were less likely to have worsened psychosocial well-being. CONCLUSION: Although more women in IMPG experienced restored PWBC after 2 years, those in AUTOG exhibited a more favorable postoperative trajectory of change in PWBC. This finding can inform clinical treatment decisions, help manage patient expectations for recovery, and develop rehabilitation interventions contributing to enhancing the postoperative quality of life for breast cancer patients.

Effectiveness of routine provision of feedback from patient-reported outcome measurements for cancer care improvement: a systematic review and meta-analysis.

BACKGROUND: Research shows that feeding back patient-reported outcome information to clinicians and/or patients could be associated with improved care processes and patient outcomes. Quantitative syntheses of intervention effects on oncology patient outcomes are lacking. OBJECTIVE: To determine the effects of patient-reported outcome measure (PROM) feedback intervention on oncology patient outcomes. DATA SOURCES: We identified relevant studies from 116 references included in our previous Cochrane review assessing the intervention for the general population. In May 2022, we conducted a systematic search in five bibliography databases using predefined keywords for additional studies published after the Cochrane review. STUDY SELECTION: We included randomized controlled trials evaluating the effects of PROM feedback intervention on processes and outcomes of care for oncology patients. DATA EXTRACTION AND SYNTHESIS: We used the meta-analytic approach to synthesize across studies measuring the same outcomes. We estimated pooled effects of the intervention on outcomes using Cohen's d for continuous data and risk ratio (RR) with a 95% confidence interval for dichotomous data. We used a descriptive approach to summarize studies which reported insufficient data for a meta-analysis. MAIN OUTCOME(S) AND MEASURES(S): Health-related quality of life (HRQL), symptoms, patient-healthcare provider communication, number of visits and hospitalizations, number of adverse events, and overall survival. RESULTS: We included 29 studies involving 7071 cancer participants. A small number of studies was available for each metanalysis (median = 3 studies, ranging from 2 to 9 studies) due to heterogeneity in the evaluation of the trials. We found that the intervention improved HRQL (Cohen's d = 0.23, 95% CI 0.11-0.34), mental functioning (Cohen's d = 0.14, 95% CI 0.02-0.26), patient-healthcare provider communication (Cohen's d = 0.41, 95% CI 0.20-0.62), and 1-year overall survival (OR = 0.64, 95% CI 0.48-0.86). The risk of bias across studies was considerable in the domains of allocation concealment, blinding, and intervention contamination. CONCLUSIONS AND RELEVANCE: Although we found evidence to support the intervention for highly relevant outcomes, our conclusions are tempered by the high risk of bias relating mainly to intervention design. PROM feedback for oncology patients may improve processes and outcomes for cancer patients but more high-quality evidence is required.

Ensuring fair, safe, and interpretable artificial intelligence-based prediction tools in a real-world oncological setting.

BACKGROUND: Cancer patients often experience treatment-related symptoms which, if uncontrolled, may require emergency department admission. We developed models identifying breast or genitourinary cancer patients at the risk of attending emergency department (ED) within 30-days and demonstrated the development, validation, and proactive approach to in-production monitoring of an artificial intelligence-based predictive model during a 3-month simulated deployment at a cancer hospital in the United States. METHODS: We used routinely-collected electronic health record data to develop our predictive models. We evaluated models including a variational autoencoder k-nearest neighbors algorithm (VAE-kNN) and model behaviors with a sample containing 84,138 observations from 28,369 patients. We assessed the model during a 77-day production period exposure to live data using a proactively monitoring process with predefined metrics. RESULTS: Performance of the VAE-kNN algorithm is exceptional (Area under the receiver-operating characteristics, AUC = 0.80) and remains stable across demographic and disease groups over the production period (AUC 0.74-0.82). We can detect issues in data feeds using our monitoring process to create immediate insights into future model performance. CONCLUSIONS: Our algorithm demonstrates exceptional performance at predicting risk of 30-day ED visits. We confirm that model outputs are equitable and stable over time using a proactive monitoring approach.

Patients with cancer often need to visit the hospital emergency department (ED), for example due to treatment side effects. Predicting these visits might help us to better manage the treatment of patients who are at risk. Here, we develop a computer-based tool to identify patients with cancer who are at risk of an unplanned ED visit within 30 days. We use health record data from over 28,000 patients who had visited a single cancer hospital in the US to create and test the model. The model performed well and was consistent across different demographic and disease groups. We monitor model behavior over time and show that it is stable. The approach we take to monitoring model performance may be a particularly useful contribution toward implementing similar predictive models in the clinic and checking that they are performing as intended.

Machine learning and patient-reported outcomes for longitudinal monitoring of disease progression in metastatic breast cancer: a multicenter, retrospective analysis.

BACKGROUND: Assessments of health-related quality of life (HRQoL) play an important role in transition to palliative care for women with metastatic breast cancer. We developed machine learning (ML) algorithms to analyse longitudinal HRQoL data and identify patients who may benefit from palliative care due to disease progression. METHODS: We recruited patients from two institutions and administered the EuroQoL Visual Analog Scale (EQ-VAS) via an online platform over a 6-month period. We trained a regularised regression algorithm using 10-fold cross-validation to determine if a patient was at high or low risk of disease progression based on changes in the EQ-VAS scores using data of one institution and validated the performance on data of the other institution. Progression-free survival (PFS) was the end-point. We conducted Kaplan-Meier and Cox regression analysis adjusted for clinical risk factors. RESULTS: Of 179 patients, 98 (54.7%) had progressive disease after a median follow-up of 14weeks. Using EQ-VAS scores collected at weeks 1-6 to predict disease progression at week 12, in the validation set (n = 63), PFS was significantly lower in the intelligent EQ-VAS high-risk versus low-risk group: median PFS 7 versus 10weeks, log-rank P < 0.038). Intelligent EQ-VAS had the strongest association with PFS (adjusted hazard ratio 2.69, 95% confidence interval 1.17-6.18, P = 0.02). CONCLUSION: ML algorithms can analyse changes in longitudinal HRQoL data to identify patients with disease progression earlier than standard follow-up methods. Intelligent EQ-VAS scores were identified as independent prognostic factor. Future studies may validate these results to remotely monitor patients.

The Development, Deployment, and Evaluation of the CLEFT-Q Computerized Adaptive Test: A Multimethods Approach Contributing to Personalized, Person-Centered Health Assessments in Plastic Surgery.

BACKGROUND: Routine use of patient-reported outcome measures (PROMs) and computerized adaptive tests (CATs) may improve care in a range of surgical conditions. However, most available CATs are neither condition-specific nor coproduced with patients and lack clinically relevant score interpretation. Recently, a PROM called the CLEFT-Q has been developed for use in the treatment of cleft lip or palate (CL/P), but the assessment burden may be limiting its uptake into clinical practice. OBJECTIVE: We aimed to develop a CAT for the CLEFT-Q, which could facilitate the uptake of the CLEFT-Q PROM internationally. We aimed to conduct this work with a novel patient-centered approach and make source code available as an open-source framework for CAT development in other surgical conditions. METHODS: CATs were developed with the Rasch measurement theory, using full-length CLEFT-Q responses collected during the CLEFT-Q field test (this included 2434 patients across 12 countries). These algorithms were validated in Monte Carlo simulations involving full-length CLEFT-Q responses collected from 536 patients. In these simulations, the CAT algorithms approximated full-length CLEFT-Q scores iteratively, using progressively fewer items from the full-length PROM. Agreement between full-length CLEFT-Q score and CAT score at different assessment lengths was measured using the Pearson correlation coefficient, root-mean-square error (RMSE), and 95% limits of agreement. CAT settings, including the number of items to be included in the final assessments, were determined in a multistakeholder workshop that included patients and health care professionals. A user interface was developed for the platform, and it was prospectively piloted in the United Kingdom and the Netherlands. Interviews were conducted with 6 patients and 4 clinicians to explore end-user experience. RESULTS: The length of all 8 CLEFT-Q scales in the International Consortium for Health Outcomes Measurement (ICHOM) Standard Set combined was reduced from 76 to 59 items, and at this length, CAT assessments reproduced full-length CLEFT-Q scores accurately (with correlations between full-length CLEFT-Q score and CAT score exceeding 0.97, and the RMSE ranging from 2 to 5 out of 100). Workshop stakeholders considered this the optimal balance between accuracy and assessment burden. The platform was perceived to improve clinical communication and facilitate shared decision-making. CONCLUSIONS: Our platform is likely to facilitate routine CLEFT-Q uptake, and this may have a positive impact on clinical care. Our free source code enables other researchers to rapidly and economically reproduce this work for other PROMs.

Machine Learning Models Using Routinely Collected Clinical Data Offer Robust and Interpretable Predictions of 90-Day Unplanned Acute Care Use for Cancer Immunotherapy Patients.

PURPOSE: Clinical management of patients receiving immune checkpoint inhibitors (ICIs) could be informed using accurate predictive tools to identify patients at risk of short-term acute care utilization (ACU). We used routinely collected data to develop and assess machine learning (ML) algorithms to predict unplanned ACU within 90 days of ICI treatment initiation. METHODS: We used aggregated electronic health record data from 7,960 patients receiving ICI treatments to train and assess eight ML algorithms. We developed the models using pre-SARS-COV-19 COVID-19 data generated between January 2016 and February 2020. We validated our algorithms using data collected between March 2020 and June 2022 (peri-COVID-19 sample). We assessed performance using area under the receiver operating characteristic curves (AUROC), sensitivity, specificity, and calibration plots. We derived intuitive explanations of predictions using variable importance and Shapley additive explanation analyses. We assessed the marginal performance of ML models compared with that of univariate and multivariate logistic regression (LR) models. RESULTS: Most algorithms significantly outperformed the univariate and multivariate LR models. The extreme gradient boosting trees (XGBT) algorithm demonstrated the best overall performance (AUROC, 0.70; sensitivity, 0.53; specificity, 0.74) on the peri-COVID-19 sample. The algorithm performance was stable across both pre- and peri-COVID-19 samples, as well as ICI regimen and cancer groups. Type of ICI agents, oxygen saturation, diastolic blood pressure, albumin level, platelet count, immature granulocytes, absolute monocyte, chloride level, red cell distribution width, and alcohol intake were the top 10 key predictors used by the XGBT algorithm. CONCLUSION: Machine learning algorithms trained using routinely collected data outperformed traditional statistical models when predicting 90-day ACU. The XGBT algorithm has the potential to identify high-ACU risk patients and enable preventive interventions to avoid ACU.

On the importance of interpretable machine learning predictions to inform clinical decision making in oncology.

Machine learning-based tools are capable of guiding individualized clinical management and decision-making by providing predictions of a patient's future health state. Through their ability to model complex nonlinear relationships, ML algorithms can often outperform traditional statistical prediction approaches, but the use of nonlinear functions can mean that ML techniques may also be less interpretable than traditional statistical methodologies. While there are benefits of intrinsic interpretability, many model-agnostic approaches now exist and can provide insight into the way in which ML systems make decisions. In this paper, we describe how different algorithms can be interpreted and introduce some techniques for interpreting complex nonlinear algorithms.

Machine learning models for 180-day mortality prediction of patients with advanced cancer using patient-reported symptom data.

PURPOSE: The objective of the current study was to develop and test the performances of different ML algorithms which were trained using patient-reported symptom severity data to predict mortality within 180 days for patients with advanced cancer. METHODS: We randomly selected 630 of 689 patients with advanced cancer at our institution who completed symptom PRO measures as part of routine care between 2009 and 2020. Using clinical, demographic, and PRO data, we trained and tested four ML algorithms: generalized regression with elastic net regularization (GLM), extreme gradient boosting (XGBoost) trees, support vector machines (SVM), and a single hidden layer neural network (NNET). We assessed the performance of algorithms individually as well as part of an unweighted voting ensemble on the hold-out testing sample. Performance was assessed using area under the receiver-operating characteristic curve (AUROC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). RESULTS: The starting cohort of 630 patients was randomly partitioned into training (n = 504) and testing (n = 126) samples. Of the four ML models, the XGBoost algorithm demonstrated the best performance for 180-day mortality prediction in testing data (AUROC = 0.69, sensitivity = 0.68, specificity = 0.62, PPV = 0.66, NPV = 0.64). Ensemble of all algorithms performed worst (AUROC = 0.65, sensitivity = 0.65, specificity = 0.62, PPV = 0.65, NPV = 0.62). Of individual PRO symptoms, shortness of breath emerged as the variable of highest impact on the XGBoost 180-mortality prediction (1-AUROC = 0.30). CONCLUSION: Our findings support ML models driven by patient-reported symptom severity as accurate predictors of short-term mortality in patients with advanced cancer, highlighting the opportunity to integrate these models prospectively into future studies of goal-concordant care.

Towards Patient-centered Decision-making in Breast Cancer Surgery: Machine Learning to Predict Individual Patient-reported Outcomes at 1-year Follow-up.

OBJECTIVE: We developed, tested, and validated machine learning algorithms to predict individual patient-reported outcomes at 1-year follow-up to facilitate individualized, patient-centered decision-making for women with breast cancer. SUMMARY OF BACKGROUND DATA: Satisfaction with breasts is a key outcome for women undergoing cancer-related mastectomy and reconstruction. Current decision-making relies on group-level evidence which may lead to suboptimal treatment recommendations for individuals. METHODS: We trained, tested, and validated 3 machine learning algorithms using data from 1921 women undergoing cancer-related mastectomy and reconstruction conducted at eleven study sites in North America from 2011 to 2016. Data from 1921 women undergoing cancer-related mastectomy and reconstruction were collected before surgery and at 1-year follow-up. Data from 10 of the 11 sites were randomly split into training and test samples (2:1 ratio) to develop and test 3 algorithms (logistic regression with elastic net penalty, extreme gradient boosting tree, and neural network) which were further validated using the additional site's data.AUC to predict clinically-significant changes in satisfaction with breasts at 1-year follow-up using the validated BREAST-Q were the outcome measures. RESULTS: The 3 algorithms performed equally well when predicting both improved or decreased satisfaction with breasts in both testing and validation datasets: For the testing dataset median accuracy = 0.81 (range 0.73-0.83), median AUC = 0.84 (range 0.78-0.85). For the validation dataset median accuracy = 0.83 (range 0.81-0.84), median AUC = 0.86 (range 0.83-0.89). CONCLUSION: Individual patient-reported outcomes can be accurately predicted using machine learning algorithms, which may facilitate individualized, patient-centered decision-making for women undergoing breast cancer treatment.

Predicting 180-day mortality for women with ovarian cancer using machine learning and patient-reported outcome data.

Contrary to national guidelines, women with ovarian cancer often receive treatment at the end of life, potentially due to the difficulty in accurately estimating prognosis. We trained machine learning algorithms to guide prognosis by predicting 180-day mortality for women with ovarian cancer using patient-reported outcomes (PRO) data. We collected data from a single academic cancer institution in the United States. Women completed biopsychosocial PRO measures every 90 days. We randomly partitioned our dataset into training and testing samples. We used synthetic minority oversampling to reduce class imbalance in the training dataset. We fitted training data to six machine learning algorithms and combined their classifications on the testing dataset into an unweighted voting ensemble. We assessed each algorithm's accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC) using testing data. We recruited 245 patients who completed 1319 PRO assessments. The final voting ensemble produced state-of-the-art results on the task of predicting 180-day mortality for ovarian cancer paitents (Accuracy = 0.79, Sensitivity = 0.71, Specificity = 0.80, AUROC = 0.76). The algorithm correctly identified 25 of the 35 women in the testing dataset who died within 180 days of assessment. Machine learning algorithms trained using PRO data offer encouraging performance in predicting whether a woman with ovarian cancer will die within 180 days. This model could be used to drive data-driven end-of-life care and address current shortcomings in care delivery. Our model demonstrates the potential of biopsychosocial PROM information to make substantial contributions to oncology prediction modeling. This model could inform clinical decision-making Future research is needed to validate these findings in a larger, more diverse sample.