Using Machine Learning to Predict Mortality in Older Patients With Cancer: Development and External Validation of the Geriatric Cancer Scoring System (GCSS)

2021 ◽  
Author(s):  
Etienne Audureau ◽  
Pierre Soubeyran ◽  
Claudia Martinez-Tapia ◽  
Carine Bellera ◽  
Sylvie Bastuji-Garin ◽  
...  
Keyword(s):  
Machine Learning ◽  
Scoring System ◽  
Older Patients ◽  
External Validation ◽  
Cancer Development ◽  
Patients With Cancer ◽  
Geriatric Cancer

scholarly journals One-Year Mortality in Older Patients with Cancer: Development and External Validation of an MNA-Based Prognostic Score

PLoS ONE ◽  
2016 ◽  
Vol 11 (2) ◽  
pp. e0148523 ◽  
Author(s):  
Isabelle Bourdel-Marchasson ◽  
Abou Diallo ◽  
Carine Bellera ◽  
Christelle Blanc-Bisson ◽  
Jessica Durrieu ◽  
...  
Keyword(s):  
Older Patients ◽  
Prognostic Score ◽  
External Validation ◽  
Cancer Development ◽  
Patients With Cancer ◽  
One Year

scholarly journals Development and Validation of an Early Mortality Risk Score for Older Patients Treated with Chemotherapy for Cancer

2021 ◽  
Vol 10 (8) ◽  
pp. 1615
Author(s):  
Jaime Feliu ◽  
Alvaro Pinto ◽  
Laura Basterretxea ◽  
Borja López-San Vicente ◽  
Irene Paredero ◽  
...  
Keyword(s):  
Older Patients ◽  
Prognostic Score ◽  
External Validation ◽  
Early Death ◽  
Early Mortality ◽  
Background Estimation ◽  
Patients With Cancer ◽  
Validation Set ◽  
Ecog Ps ◽  
Development And Validation

Background: Estimation of life expectancy in older patients is relevant to select the best treatment strategy. We aimed to develop and validate a score to predict early mortality in older patients with cancer. Patients and Methods: A total of 749 patients over 70 years starting new chemotherapy regimens were prospectively included. A prechemotherapy assessment that included sociodemographic variables, tumor/treatment variables, and geriatric assessment variables was performed. Association between these factors and early death was examined using multivariable logistic regression. Score points were assigned to each risk factor. External validation was performed on an independent cohort. Results: In the training cohort, the independent predictors of 6-month mortality were metastatic stage (OR 4.8, 95% CI [2.4–9.6]), ECOG-PS 2 (OR 2.3, 95% CI [1.1–5.2]), ADL ≤ 5 (OR 1.7, 95% CI [1.1–3.5]), serum albumin levels ≤ 3.5 g/dL (OR 3.4, 95% CI [1.7–6.6]), BMI < 23 kg/m2 (OR 2.5, 95% CI [1.3–4.9]), and hemoglobin levels < 11 g/dL (OR 2.4, 95% CI (1.2–4.7)). With these results, we built a prognostic score. The area under the ROC curve was 0.78 (95% CI, 0.73 to 0.84), and in the validation set, it was 0.73 (95% CI: 0.67–0.79). Conclusions: This simple and highly accurate tool can help physicians making decisions in elderly patients with cancer who are planned to initiate chemotherapy treatment.

scholarly journals Comprehensive geriatric assessment in older patients with cancer: an external validation of the Multidimensional Prognostic Index in a French prospective cohort study

2019 ◽  
Author(s):  
Chunyun Hu ◽  
Marc Paccalin ◽  
Simon Valero ◽  
Amelie Jamet ◽  
Thomas Brunet ◽  
...  
Keyword(s):  
Geriatric Assessment ◽  
Comprehensive Geriatric Assessment ◽  
Older Patients ◽  
External Validation ◽  
Prognostic Index ◽  
Risk Groups ◽  
University Hospital ◽  
Patients With Cancer ◽  
Cox Proportional Hazard Regression ◽  
One Year

Abstract Background: Older patients with cancer require specific and individualized management. The Multidimensional Prognostic Index (MPI) based on the Comprehensive Geriatric Assessment (CGA) has shown a predictive interest in terms of mortality.Methods: From 2015 to 2017, consecutive patients ≥75 years old with cancer in Poitiers University Hospital referred to an oncogeriatric consultation. Patients underwent CGA with MPI that is categorized into three risk groups of mortality at one year.Results: Overall, 433 patients were included (women 42%; mean age 82.8±4.8 years). Most common tumor sites were prostate (23%), skin (17%), colorectum (15%) and breast (12%); 29% patients had a metastatic disease; 231 patients (53%) belonged to "MPI-1" group, 172 (40%) to "MPI-2" group and 30 patients (7%) were classified in "MPI-3" group. One-year mortality rate was 32% (23% in MPI-1, 41% in MPI-2 and 53% in MPI-3, p=0.024). All domains of MPI except cognition and living status were significantly associated with mortality at one-year, as well as tumor sites and metastatic status. Cox proportional hazard regression analysis, adjusted on age, gender, tumor sites and metastatic status, validated MPI as being associated with a higher mortality risk (p<0.0001). The prognostic value of MPI was confirmed by the area under the ROC curve at 0.826 (P <0.0001).Conclusion: Our study confirmed the predictive value of MPI for one-year mortality in older patients with cancer. This practical prognostic tool may help to optimize the management of these vulnerable patients.

scholarly journals Development and validation of a new scoring system for prognostic prediction of community-acquired pneumonia in older adults

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Masahiro Shirata ◽  
Isao Ito ◽  
Tadashi Ishida ◽  
Hiromasa Tachibana ◽  
Naoya Tanabe ◽  
...  
Keyword(s):  
Confidence Interval ◽  
Scoring System ◽  
Older Patients ◽  
Characteristic Curve ◽  
External Validation ◽  
Albumin Level ◽  
Community Acquired Pneumonia ◽  
Derivation Cohort ◽  
Prognostic Prediction ◽  
New Scoring

AbstractThe discriminative power of CURB-65 for mortality in community-acquired pneumonia (CAP) is suspected to decrease with age. However, a useful prognostic prediction model for older patients with CAP has not been established. This study aimed to develop and validate a new scoring system for predicting mortality in older patients with CAP. We recruited two prospective cohorts including patients aged ≥ 65 years and hospitalized with CAP. In the derivation (n = 872) and validation cohorts (n = 1,158), the average age was 82.0 and 80.6 years and the 30-day mortality rate was 7.6% (n = 66) and 7.4% (n = 86), respectively. A new scoring system was developed based on factors associated with 30-day mortality, identified by multivariate analysis in the derivation cohort. This scoring system named CHUBA comprised five variables: confusion, hypoxemia (SpO2 ≤ 90% or PaO2 ≤ 60 mmHg), blood urea nitrogen ≥ 30 mg/dL, bedridden state, and serum albumin level ≤ 3.0 g/dL. With regard to 30-day mortality, the area under the receiver operating characteristic curve for CURB-65 and CHUBA was 0.672 (95% confidence interval, 0.607–0.732) and 0.809 (95% confidence interval, 0.751–0.856; P < 0.001), respectively. The effectiveness of CHUBA was statistically confirmed in the external validation cohort. In conclusion, a simpler novel scoring system, CHUBA, was established for predicting mortality in older patients with CAP.

Development and validation of an early death risk score for older patients treated with chemotherapy for cancer.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. 12030-12030
Author(s):  
Jaime Feliu Batlle ◽  
Alvaro Pinto ◽  
Laura Basterretxea ◽  
Irene Paredero Pérez ◽  
Elisenda Llabres ◽  
...  
Keyword(s):  
Elderly Patients ◽  
Risk Score ◽  
Older Patients ◽  
Validation Cohort ◽  
External Validation ◽  
Early Death ◽  
Chemotherapy Treatment ◽  
Training Cohort ◽  
Death Risk ◽  
Patients With Cancer

12030 Background: Determining life expectancy in older patients is needed to select the best treatment strategy. We aimed to develop and validate a score to predict early death risk ( < 6 months) in elderly patients with cancer that are planned to initiate chemotherapy treatment. Methods: Patients over 70 years starting new chemotherapy regimens were prospectively included in a multicenter study. A pre-chemotherapy assessment that included sociodemographics, tumor/treatment variables, and geriatric assessment variables, was performed. Association between these factors and early death was examined by using multivariate logistic regression. Score points were assigned to each risk factor based on their b coefficient. We validated the risk score with an external validation cohort of 206 patients. Results: Three hundred forty two patients were included in the training cohort. The independent predictors for early death were metastasic cancers (odds ratio [OR] 4.8, 95% confidence interval [CI], [2.4-9.6]), ECOG performance status (OR 2.3, 95% CI:1.084-5.232), ADL (OR 1.7, 95% CI:1.08-3.5), serum albumin levels (3.3, 95% CI: 1.6-6.6), BMI (OR 2.4, 95% CI:1,2-4.8), serum GGT levels (OR 1.5, 95% CI:1.05-1.8) and hemoglobin levels (OR 2.3, 95% CI:1.2-4.6). With these results, a score was to stratify patients regarding their risk of early death: low (0 to 2 points; 5%), intermediate (3 to 5 points; 19%) or high (6 to 14 points; 50%) (p < 0.001). The area under the curve of the receiver-operating characteristic (ROC) curve was 0.79 for the training cohort (95% CI, 0.74 to 0.85), and 0.70 (95% CI: 0.60-0.80) for the validation cohort (difference between cohorts not statistically different). Conclusions: We developed a highly accurate tool that uses basic clinical and analytical information to predict the probability of early death in elderly patients with cancer that are planned to initiate chemotherapy treatment. This tool can help physicians in decision making for this population of patients.

scholarly journals A Novel Prognostic Scoring System of Intrahepatic Cholangiocarcinoma With Machine Learning Basing on Real-World Data

2021 ◽  
Vol 10 ◽  
Author(s):  
Zhizhen Li ◽  
Lei Yuan ◽  
Chen Zhang ◽  
Jiaxing Sun ◽  
Zeyuan Wang ◽  
...  
Keyword(s):  
Machine Learning ◽  
Scoring System ◽  
Cox Regression ◽  
Validation Cohort ◽  
External Validation ◽  
Machine Learning Techniques ◽  
Internal Validation ◽  
Learning Techniques ◽  
Staging Systems ◽  
8Th Edition

Background and ObjectivesCurrently, the prognostic performance of the staging systems proposed by the 8th edition of the American Joint Committee on Cancer (AJCC 8th) and the Liver Cancer Study Group of Japan (LCSGJ) in resectable intrahepatic cholangiocarcinoma (ICC) remains controversial. The aim of this study was to use machine learning techniques to modify existing ICC staging strategies based on clinical data and to demonstrate the accuracy and discrimination capacity in prognostic prediction.Patients and MethodsThis is a retrospective study based on 1,390 patients who underwent surgical resection for ICC at Eastern Hepatobiliary Surgery Hospital from 2007 to 2015. External validation was performed for patients from 2015 to 2017. The ensemble of three machine learning algorithms was used to select the most important prognostic factors and stepwise Cox regression was employed to derive a modified scoring system. The discriminative ability and predictive accuracy were assessed using the Concordance Index (C-index) and Brier Score (BS). The results were externally validated through a cohort of 42 patients operated on from the same institution.ResultsSix independent prognosis factors were selected and incorporated in the modified scoring system, including carcinoembryonic antigen, carbohydrate antigen 19-9, alpha-fetoprotein, prealbumin, T and N of ICC staging category in 8th edition of AJCC. The proposed scoring system showed a more favorable discriminatory ability and model performance than the AJCC 8th and LCSGJ staging systems, with a higher C-index of 0.693 (95% CI, 0.663–0.723) in the internal validation cohort and 0.671 (95% CI, 0.602–0.740) in the external validation cohort, which was then confirmed with lower BS (0.103 in internal validation cohort and 0.169 in external validation cohort). Meanwhile, machine learning techniques for variable selection together with stepwise Cox regression for survival analysis shows a better prognostic accuracy than using stepwise Cox regression method only.ConclusionsThis study put forward a modified ICC scoring system based on prognosis factors selection incorporated with machine learning, for individualized prognosis evaluation in patients with ICC.

Using machine learning to predict mortality in older patients with cancer: Decision tree and random forest analyses from the ELCAPA and ONCODAGE prospective cohorts.

2019 ◽  
Vol 37 (15_suppl) ◽  
pp. 11516-11516
Author(s):  
Etienne Audureau ◽  
Pierre-Louis Soubeyran ◽  
Claudia Martinez-Tapia ◽  
Carine A. Bellera ◽  
Sylvie Bastuji-Garin ◽  
...  
Keyword(s):  
Machine Learning ◽  
Random Forest ◽  
Decision Tree ◽  
Older Patients ◽  
Cox Model ◽  
Biological Data ◽  
Tumor Site ◽  
Patients With Cancer ◽  
Complex Interactions ◽  
Prospective Cohorts

11516 Background: Accurate prognosis is crucial to decision making in oncology, but remains challenging in older patients due to the heterogeneity of this population and the lack of ability of current models to capture complex interactions between oncological and geriatric predictors. We aimed to develop new predictive algorithms based on machine learning to refine individualized prognosis in older patients with cancer. Methods: Data were collected from 3409 patients ≥70 years referred to geriatric oncology clinics for completion of a geriatric assessment (GA), including 2012 and 1397 patients from the ELCAPA (training set) and ONCODAGE (validation set) French prospective cohorts, respectively. Candidate predictors included baseline oncological and geriatric parameters, G-8 score and routine biological data (CRP/albumin ratio). Prognostic models for 12-months mortality were built using Cox regression model, single decision tree (DT) and random survival forest (RSF). Models performance was compared based on externally validated Harrell’s C-indexes. Results: During the 1-year study period, 875 (43%) and 219 (16%) patients died in the training and validation sets, respectively (mean age: 81±6 / 78±5, women 47% / 70%, metastasis 50% / 34%). Cox model identified 9 independent predictors of mortality: tumor site/metastatic status, anticancer treatment, weight loss > 3kg, drugs > 5, renal failure, increased CRP/Albumin, ECOG-PS≥2, ADL≤5 and altered TGUG. DT identified more complex combinations between features, yielding 16 patient groups with highly differentiated survival, notably depending on the G-8 ( < 10 vs. ≥10 as the root node). RFS had the highest C-index (0.86 [RFS], 0.82 [Cox], 0.81 [DT]), identifying the G-8, CRP/albumin and tumor site/metastasis as the most important features. Conclusions: While Cox modeling confirmed known independent prognostic factors, DT revealed more complex interactions between them and random forest achieved superior prognostic performance by better capturing patient’s complexity. The latter model has been implemented into an interactive web interface for easy and direct use in clinical practice. Clinical trial information: NCT02884375.

Sign in / Sign up

Export Citation Format

Share Document