JCO Clinical Cancer Informatics最新文献

{"title":"Assessing Algorithmic Fairness With a Multimodal Artificial Intelligence Model in Men of African and Non-African Origin on NRG Oncology Prostate Cancer Phase III Trials.","authors":"Mack Roach, Jingbin Zhang, Osama Mohamad, Douwe van der Wal, Jeffry P Simko, Sandy DeVries, Huei-Chung Huang, Songwan Joun, Edward M Schaeffer, Todd M Morgan, Jessica Keim-Malpass, Emmalyn Chen, Rikiya Yamashita, Jedidiah M Monson, Farah Naz, James Wallace, Jean-Paul Bahary, Derek Wilke, Sonny Batra, Gregory B Biedermann, Sergio Faria, Lindsay Hwang, Howard M Sandler, Daniel E Spratt, Stephanie L Pugh, Andre Esteva, Phuoc T Tran, Felix Y Feng","doi":"10.1200/CCI-24-00284","DOIUrl":"https://doi.org/10.1200/CCI-24-00284","url":null,"abstract":"<p><strong>Purpose: </strong>Artificial intelligence (AI) tools could improve clinical decision making or exacerbate inequities because of bias. African American (AA) men reportedly have a worse prognosis for prostate cancer (PCa) and are underrepresented in the development genomic biomarkers. We assess the generalizability of tools developed using a multimodal AI (MMAI) deep learning system using digital histopathology and clinical data from NRG/Radiation Therapy Oncology Group PCa trials across racial subgroups.</p><p><strong>Methods: </strong>In total, 5,708 patients from five randomized phase III trials were included. Two MMAI algorithms were evaluated: (1) the distant metastasis (DM) MMAI model optimized to predict risk of DM, and (2) the PCa-specific mortality (PCSM) MMAI model optimized to focus on prediction death in the presence of DM (DDM). The prognostic performance of the MMAI algorithms was evaluated in AA and non-AA subgroups using time to DM (primary end point) and time to DDM (secondary end point). Exploratory end points included time to biochemical failure and overall survival with Fine-Gray or Cox proportional hazards models. Cumulative incidence estimates were computed for time-to-event end points and compared using Gray's test.</p><p><strong>Results: </strong>There were 948 (16.6%) AA patients, 4,731 non-AA patients (82.9%), and 29 (0.5%) patients with unknown or missing race status. The DM-MMAI algorithm showed a strong prognostic signal for DM in the AA (subdistribution hazard ratio [sHR], 1.2 [95% CI, 1.0 to 1.3]; <i>P</i> = .007) and non-AA subgroups (sHR, 1.4 [95% CI, 1.3 to 1.5]; <i>P</i> < .001). Similarly, the PCSM-MMAI score showed a strong prognostic signal for DDM in both AA (sHR, 1.3 [95% CI, 1.1 to 1.5]; <i>P</i> = .001) and non-AA subgroups (sHR, 1.5 [95% CI, 1.4 to 1.6]; <i>P</i> < .001), with similar distributions of risk.</p><p><strong>Conclusion: </strong>Using cooperative group data sets with a racially diverse population, the MMAI algorithm performed well across racial subgroups without evidence of algorithmic bias.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400284"},"PeriodicalIF":3.3,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144065285","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Performance of Two-Phase Designs for the Time-to-Event Outcome and a Case Study Assessing the Relapse Risk Associated With B-ALL Subtypes.","authors":"Wenan Chen, Ti-Cheng Chang, Karen R Rabin, Elizabeth A Raetz, Meenakshi Devidas, Stephen P Hunger, Nilsa C Ramirez, Charles G Mullighan, Mignon L Loh, Gang Wu","doi":"10.1200/CCI-24-00223","DOIUrl":"https://doi.org/10.1200/CCI-24-00223","url":null,"abstract":"<p><strong>Purpose: </strong>To reduce costs in genomic studies of time-to-event phenotypes like survival, researchers often sequence a subset of samples from a larger cohort. This process usually involves two phases: first, collecting inexpensive variables from all samples, and second, selecting a subset for expensive measurements, for example, sequencing-based biomarkers. Common two-phase designs include nested case-control and case-cohort designs. Additional designs include sampling subjects based on follow-up time, like extreme case-control designs. Recently an optimal two-phase design using a maximum likelihood-based method was proposed, which could accommodate arbitrary sample selection in the second phase. However, direct comparisons of this optimal design with others in terms of power and computational cost is lacking.</p><p><strong>Methods: </strong>This study performs a direct evaluation of typical two-phase designs, including Tao's optimal design, on type I error, power, effect size estimation, and computational time, using both simulated and real data sets.</p><p><strong>Results: </strong>Results show that the optimal design had the highest power and accurate effect size estimation under the Cox regression model. Surprisingly, logistic regression achieved similar power with much lower computational cost than a more sophisticated method. The study further applied these methods to the MP2PRT study, reporting hazard ratios of cancer subtypes on relapse risk.</p><p><strong>Conclusion: </strong>Recommendations for selecting two-phase designs and analysis methods are regarding power, bias of estimated effect size, and computational time.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400223"},"PeriodicalIF":3.3,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12052069/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143998303","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"User-Centered Design and Comparison of Two Electronic Health Record Tools to Support the Ordering of Crisantaspase Recombinant Chemotherapy.","authors":"Renee Potashner, Tracey Taylor, Ally Sarna, Marilyn Cooper, Siron Thayaparan, Lillian Sung, Karim Jessa, Adam P Yan","doi":"10.1200/CCI-24-00306","DOIUrl":"https://doi.org/10.1200/CCI-24-00306","url":null,"abstract":"<p><strong>Purpose: </strong>Chemotherapy ordering errors can have serious safety implications in pediatric oncology. Computerized provider order entry systems may reduce chemotherapy ordering errors. Crisantaspase recombinant (crisantaspase) is a chemotherapy drug used in pediatric leukemia that poses significant safety risk when ordering because of complex dosing and monitoring. The aim of this study was to compare errors, satisfaction, and efficiency between two approaches to ordering in our electronic health record: namely, the standard treatment plan order group (OG) and a novel supportive care plan (SCP).</p><p><strong>Methods: </strong>We recruited oncology providers and nurses at an academic pediatric institution. Providers were asked to complete two simulated chemotherapy ordering sessions using the treatment plan OG and the SCP. Order entry errors were assessed in seven domains, and the total number of order entry errors was calculated. Satisfaction was assessed using a five-point Likert scale, and satisfaction was defined as answering \"Agree\" or \"Strongly Agree\" to all five satisfaction questions. Efficiency was compared by measuring the time to complete the task. Errors, satisfaction, and efficiency were compared between the two tools.</p><p><strong>Results: </strong>We enrolled 14 providers and five nurses. The proportion of chemotherapy ordering errors was significantly lower with the SCP (5 of 98, 5.1%) compared with the treatment plan OG (11 of 98, 11.2%; <i>P</i> < .01). The SCP significantly improved provider efficiency, reducing the time taken to complete order entry from 16.3 minutes with the OG to 7.7 minutes with the SCP (mean difference, 8.6 minutes; <i>P</i> < .001). Provider satisfaction was significantly higher with the SCP (12 of 14, 85.7%) compared with the treatment plan OG (2 of 14, 14.2%; <i>P</i> < .001).</p><p><strong>Conclusion: </strong>Use of a novel SCP instead of a tradition treatment plan OG improved provider efficiency and satisfaction while decreasing order entry errors. Thoughtful design and usability testing of chemotherapy order tools is needed to maximize their utility.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400306"},"PeriodicalIF":3.3,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144081995","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Telehealth and Emergency Department Use Among Commercially Insured, Medicaid, and Medicare Patients Receiving Systemic Cancer Therapy in Washington State After COVID-19.","authors":"Scott D Ramsey, Qin Sun, Catherine R Fedorenko, Li Li, Laura E Panattoni, Karma L Kreizenbeck, Veena Shankaran","doi":"10.1200/CCI-24-00217","DOIUrl":"https://doi.org/10.1200/CCI-24-00217","url":null,"abstract":"<p><strong>Purpose: </strong>In oncology, telehealth services were adopted as a means of mitigating the risk of COVID-19 transmission. We hypothesized that Medicaid enrollees would have less access to telehealth than commercially insured or Medicare enrollees during the pandemic, resulting in higher rates of emergency department (ED) visits during systemic cancer treatment.</p><p><strong>Methods: </strong>Linking Washington State SEER records with commercial, Medicaid, and Medicare records, we evaluated adults with new solid tumor malignancies who received initial systemic treatment before the COVID-19 pandemic (January 1, 2017-December 31, 2019) and after the pandemic (March 1, 2020-November 30, 2021). Poisson and logistic regressions were used to evaluate differences in the number of office visits, telehealth visits, and ED visits in the 3 months after starting systemic anticancer treatment between insurance groups before versus after the pandemic.</p><p><strong>Results: </strong>Among 2,936 commercial, 2,039 Medicaid, and 7,333 Medicare enrollees who met inclusion criteria, office-based visits fell substantially for all groups during the COVID-19 period. Medicare enrollees had fewer telehealth visits while Medicaid had more telehealth visits, compared with commercial enrollees. ED visits declined for all patients, but there were no differences between insurance groups.</p><p><strong>Conclusion: </strong>In Washington State, COVID-19 resulted in a substantial decrease in office-based visits, with an accompanying increase in telehealth visits partially offsetting the difference in overall access to care. ED visit rates fell substantially, without differences between insurance groups.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400217"},"PeriodicalIF":3.3,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144121426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Creating a Proxy for Baseline Eastern Cooperative Oncology Group Performance Status in Electronic Health Records for Comparative Effectiveness Research in Advanced Non-Small Cell Lung Cancer.","authors":"Michael Johnson, Peining Tao, Mehmet Burcu, John Kang, Richard Baumgartner, Junshui Ma, Vladimir Svetnik","doi":"10.1200/CCI-24-00185","DOIUrl":"10.1200/CCI-24-00185","url":null,"abstract":"<p><strong>Purpose: </strong>Eastern Cooperative Oncology Group performance status (ECOG PS) is a key confounder in comparative effectiveness research, predicting treatment and survival, but is often incomplete in electronic health records (EHRs). Imputation on the basis of classification metrics alone may introduce differences in survival between patients with known and imputed ECOG PS, complicating comparative effectiveness research. We developed an approach to impute ECOG PS so that those with known and imputed ECOG PS are indistinguishable in their survival, reducing potential biases introduced by the imputation.</p><p><strong>Methods: </strong>We analyzed deidentified data from an EHR-derived database for patients with advanced non-small cell lung cancer (aNSCLC) at their first line of treatment. Our novel imputation method involved (1) sample-splitting patients with known ECOG PS into modeling and thresholding data sets, (2) developing a predictive model of ECOG PS, (3) determining an optimal threshold aligning clinical outcomes, where a choice of outcome metric may depend on the use case, and (4) applying the model and threshold to impute missing ECOG PS. We evaluated the approach using binary classification metrics and alignment of survival metrics between observed and imputed ECOG PS.</p><p><strong>Results: </strong>Of 62,101 patients, 13,297 (21%) had missing ECOG PS at the start of their first treatment. Our method achieved similar or better performance in accuracy (73.3%), sensitivity (42.4%), and specificity (81%) compared with other techniques, with smaller survival metric differences between observed and imputed ECOG PS, with differences of 0.07 in hazard ratio, -0.36 months in median survival for good ECOG PS (<2), and -0.39 months for poor ECOG PS (≥2).</p><p><strong>Conclusion: </strong>Our imputed ECOG PS aligning clinical outcomes enhanced the use of real-world EHR data of patients with aNSCLC for comparative effectiveness research.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400185"},"PeriodicalIF":3.3,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143781526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Clinician and Patient Perspectives on a Patient-Facing Online Breast Cancer Symptom Visualization Tool.","authors":"Gillian Gresham, Michael Luu, N Lynn Henry, Tyra Nguyen, Katherine Barnhill, Greg Yothers, Sungjin Kim, Andre Rogatko, Deanna J Attai, Mourad Tighiouart, Ron D Hays, Patricia A Ganz","doi":"10.1200/CCI.24.00109","DOIUrl":"10.1200/CCI.24.00109","url":null,"abstract":"<p><strong>Purpose: </strong>Endocrine treatments for patients with hormone-sensitive breast cancer are associated with significant side effects that can negatively affect health-related quality of life and result in treatment discontinuation. The objective of this qualitative study was to obtain feedback from stakeholder clinicians and patients about an online interactive tool that was designed to provide information and visualizations of breast cancer symptoms.</p><p><strong>Methods: </strong>The online Breast Cancer Symptom Explorer tool was developed to allow patients to visualize trajectories for common symptoms associated with tamoxifen and anastrozole using symptom data from the NSABP B35 breast cancer clinical trial. To refine the tool, virtual focus groups were conducted among oncology clinicians and women with a history of breast cancer who had received treatment with an aromatase inhibitor or tamoxifen, seeking feedback on the tool and its potential usefulness. Discussions took place using a secure web-conferencing platform following a semi-structured interview guide. Focus groups were audio-recorded, transcribed, and analyzed using reflexive thematic analysis.</p><p><strong>Results: </strong>Nine focus groups were conducted (n = 21 participants: eight clinicians and 13 patients). Key benefits and barriers to tool use emerged from the discussions. Both patients and oncologists valued the ability to engage with the tool and visualize symptoms over time. They indicated that ideal settings for its use would be at home before treatment initiation. Combinations of graphical representations with text were perceived to be most effective in communicating symptoms. Key barriers identified included concerns about accessibility to the tool and digital literacy, with recommendations to simplify the text and provide health literacy support to enhance its clinical utility in the future.</p><p><strong>Conclusion: </strong>Clinician and patient involvement was critical for refinement of the breast cancer symptom explorer and provided insights into its future use and evaluation of the tool in clinical decision making.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400109"},"PeriodicalIF":3.3,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143784443","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Decoding Recurrence in Early-Stage and Locoregionally Advanced Non-Small Cell Lung Cancer: Insights From Electronic Health Records and Natural Language Processing.","authors":"Kyeryoung Lee, Zongzhi Liu, Qing Huang, David Corrigan, Iftekhar Kalsekar, Tomi Jun, Gustavo Stolovitzky, William K Oh, Ravi Rajaram, Xiaoyan Wang","doi":"10.1200/CCI-24-00227","DOIUrl":"https://doi.org/10.1200/CCI-24-00227","url":null,"abstract":"<p><strong>Purpose: </strong>Recurrences after curative resection in early-stage and locoregionally advanced non-small cell lung cancer (NSCLC) are common, necessitating a nuanced understanding of associated risk factors. This study aimed to establish a natural language processing (NLP) system to efficiently curate recurrence data in NSCLC and analyze risk factors longitudinally.</p><p><strong>Patients and methods: </strong>Electronic health records of 6,351 patients with NSCLC with >700,000 notes were obtained from Mount Sinai's data sets. A deep learning-based customized NLP system was developed to identify cohorts experiencing recurrence. Recurrence types and rates over time were stratified by various clinical features. Cohort description analysis, Kaplan-Meier analysis for overall recurrence-free survival (RFS) and distant metastasis-free survival (DMFS), and Cox proportional hazards analysis were performed.</p><p><strong>Results: </strong>Of 1,295 patients with stage I-IIIA NSCLC with surgical resections, 336 patients (25.9%) experienced recurrence, as identified through NLP. The NLP system achieved a precision of 94.3%, a recall of 93%, and an F1 score of 93.5. Among 336 patients, 52.4% had local/regional recurrences, 44% distant metastases, and 3.6% unknown recurrence. RFS rates at years 1-5 were 93%, 81%, 73%, 67%, and 61%, respectively (96%, 89%, 84%, 80%, and 75% for distant metastasis). Stage-specific RFS rates at year 5 were 73% (IA), 62% (IB), 47% (IIA), 46% (IIB), and 20% (IIIA). Stage IB patients had a significantly higher likelihood of recurrence versus stage IA (adjusted hazard ratio [aHR], 1.63; <i>P</i> = .02). The RFS was lower in patients with clinically significant <i>TP53</i> alteration (<i>v</i> <i>TP53</i>-negative or unknown significance), affecting overall RFS (aHR, 1.89; <i>P</i> = .007) and DMFS (aHR, 2.47; <i>P</i> = .009) among stage IA/IB patients.</p><p><strong>Conclusion: </strong>Our scalable NLP system enabled us to generate real-world insights into NSCLC recurrences, paving the way for predictive models for preventing, diagnosing, and treating NSCLC recurrence.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400227"},"PeriodicalIF":3.3,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12011440/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143994364","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predictive Model of Objective Response to Nivolumab Monotherapy for Advanced Renal Cell Carcinoma by Machine Learning Using Genetic and Clinical Data: The SNiP-RCC Study.","authors":"Masaki Shiota, Shota Nemoto, Ryo Ikegami, Tokiyoshi Tanegashima, Leandro Blas, Hideaki Miyake, Masayuki Takahashi, Mototsugu Oya, Norihiko Tsuchiya, Naoya Masumori, Keita Kobayashi, Wataru Obara, Nobuo Shinohara, Kiyohide Fujimoto, Masahiro Nozawa, Kojiro Ohba, Chikara Ohyama, Katsuyoshi Hashine, Shusuke Akamatsu, Takanobu Motoshima, Koji Mita, Momokazu Gotoh, Shuichi Tatarano, Masato Fujisawa, Yoshihiko Tomita, Shoichiro Mukai, Keiichi Ito, Masatoshi Eto","doi":"10.1200/CCI-24-00167","DOIUrl":"https://doi.org/10.1200/CCI-24-00167","url":null,"abstract":"<p><strong>Purpose: </strong>Anti-PD-1 antibodies are widely used for cancer treatment, including in advanced renal cell carcinoma (RCC). However, the therapeutic response varies among patients. This study aimed to predict tumor response to nivolumab anti-PD-1 antibody treatment for advanced RCC by integrating genetic and clinical data using machine learning (ML).</p><p><strong>Methods: </strong>Clinical and single-nucleotide polymorphism (SNP) data obtained in the SNPs in nivolumab PD-1 inhibitor for RCC study, which enrolled Japanese patients treated with nivolumab monotherapy for advanced clear cell RCC, were used. A point-wise linear (PWL) algorithm, logistic regression with elastic-net regularization, and eXtreme Gradient Boosting were used in this study. AUC values for objective response and C-indices for progression-free survival (PFS) were calculated to evaluate the utility of the models.</p><p><strong>Results: </strong>Among the three ML algorithms, the AUC values to predict objective response were highest for the PWL algorithm among all the data sets. Three predictive models (clinical model, small SNP model, and large SNP model) were created by the PWL algorithm using the clinical data alone and using eight and 49 SNPs in addition to the clinical data. C-indices for PFS by the clinical model, small SNP model, and large SNP model were 0.522, 0.600, and 0.635, respectively.</p><p><strong>Conclusion: </strong>The results demonstrated that the SNP models created by ML produced excellent predictions of tumor response to nivolumab monotherapy for advanced clear cell RCC and will be helpful in treatment decisions.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400167"},"PeriodicalIF":3.3,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144008427","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of a Machine Learning Algorithm to Predict Abnormalities in Serum Phosphate in a Large Oncology Cohort.","authors":"Lauren A Scanlon, Phillip J Monaghan, Safwaan Adam","doi":"10.1200/CCI-24-00312","DOIUrl":"https://doi.org/10.1200/CCI-24-00312","url":null,"abstract":"<p><strong>Purpose: </strong>Serum phosphate is commonly measured in oncology patients because of the relationship between oncologic conditions and treatments with abnormal phosphate. All patients attending our institution, a large specialist oncology center, have a standardized order set (SOS) measured. This consists of 15 biochemical tests, including serum phosphate. Our aim was to understand if abnormalities in serum phosphate could be predicted, using a machine learning algorithm (MLA) by other interrelated variables in the SOS.</p><p><strong>Methods: </strong>We trained an XGBoost MLA implemented in Python to predict occurrence of abnormal phosphate (<0.5 or >1.78 mmol/L) from other results in the SOS. To train and test this algorithm, we used 481,150 test results for 45,174 patients on blood tests between January 2019 and December 2021, with 5,897 abnormal results.</p><p><strong>Results: </strong>This model was trained and tested on a 70%/30% split (train/test result cohort), achieving an area under the receiver operator curve on the test set of 0.866 (95% CI, 0.857 to 0.875). Assigning a threshold for predictions so the model achieves a sensitivity of 0.924 and a specificity of 0.530 and only performing a phosphate test for results above this threshold, the number of phosphate tests would be reduced from 142,647 to 67,873 in this test set, capturing 1,586 of the total 1,716 abnormal results with a small risk (<0.1%) of missing an abnormal result. The model was further validated on a separate validation cohort between January 2022 and December 2023, achieving similar levels of performance.</p><p><strong>Conclusion: </strong>A MLA to optimize testing of phosphate has been developed with high sensitivity. Its application in routine care might result in cost-savings and health care efficiencies. The methodology used to develop our MLA model can be applied to other settings where interrelated variables are measured in SOS.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400312"},"PeriodicalIF":3.3,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144057267","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine Learning Models of Early Longitudinal Toxicity Trajectories Predict Cetuximab Concentration and Metastatic Colorectal Cancer Survival in the Canadian Cancer Trials Group/AGITG CO.17/20 Trials.","authors":"Danielle Lilly Nicholls, Maria C Xu, Luna Zhan, Divya Sharma, Katrina Hueniken, Kaitlyn Chiasson, Mary Wahba, M Catherine Brown, Benjamin Grant, Jeremy Shapiro, Christos S Karapetis, John Simes, Derek Jonker, Dongsheng Tu, Christopher O'Callaghan, Eric Chen, Geoffrey Liu","doi":"10.1200/CCI.24.00114","DOIUrl":"https://doi.org/10.1200/CCI.24.00114","url":null,"abstract":"<p><strong>Purpose: </strong>Cetuximab (CET), targeting the epidermal growth factor receptor, is a systemic treatment option for patients with colorectal cancer. One known predictive factor for CET efficacy is the presence of CET-related rash; other putative toxicity factors include fatigue and nausea. Analysis of early CET-associated toxicities may reveal patient subpopulations that clinically benefit from long-term CET treatment.</p><p><strong>Methods: </strong>We analyzed data from CO.20 (ClinicalTrials.gov identifier: NCT00640471) trial arms, CET + brivanib alaninate (BRIV) (n = 376) and CET + placebo (n = 374), and CO.17 (ClinicalTrials.gov identifier: NCT00079066) trial arms, CET (+best supportive care [BSC]; n = 287) and BSC only (n = 285). Patients were clustered into subpopulations using KmL3D, a machine learning method, to analyze 14 joint longitudinal toxicity trajectories from weeks 0 to 8 of treatment. Landmark survival analyses were performed from 8 weeks after treatment initiation. Regression analyses assessed the relationship between subpopulations and plasma CET concentrations. Three supervised machine learning models were developed to assign patients in the CO.20-CET trial arm into subpopulations, which were then validated using CO.20-CET-BRIV and CO.17-CET trial arm data.</p><p><strong>Results: </strong>Joint longitudinal toxicity clustering revealed dichotomous high- and low-toxicity clusters, with all CET-containing arms showing consistent toxicity trajectories and characteristics. High-toxicity clusters were associated with male predilection, fewer metastatic sites, fewer colon-only primaries, and higher body mass indices. In CO.20 trial samples, higher toxicity clusters were associated with improved overall survival and progression-free survival outcomes (adjusted hazard ratios ranging from 2.21 to 4.36) and higher CET concentrations (<i>P</i> = .003). The random forest predictive model performed the best, with an AUC of 0.981 (0.963-0.999).</p><p><strong>Conclusion: </strong>We used an innovative machine learning approach to analyze longitudinal joint drug toxicities, demonstrating their role in predicting patient outcomes through a putative pharmacokinetic mechanism.</p>","PeriodicalId":51626,"journal":{"name":"JCO Clinical Cancer Informatics","volume":"9 ","pages":"e2400114"},"PeriodicalIF":3.3,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144052649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}