Machine learning using clinical data at baseline predicts the efficacy of vedolizumab at week 22 in patients with ulcerative colitis

AbstractPredicting the response of patients with ulcerative colitis (UC) to a biologic such as vedolizumab (VDZ) before administration is an unmet need for optimizing individual patient treatment. We hypothesized that the machine-learning approach with daily clinical information can be a new, promising strategy for developing a drug-efficacy prediction tool. Random forest with grid search and cross-validation was employed in Cohort 1 to determine the contribution of clinical features at baseline (week 0) to steroid-free clinical remission (SFCR) with VDZ at week 22. Among 49 clinical features including sex, age, height, body weight, BMI, disease duration/phenotype, treatment history, clinical activity, endoscopic activity, and blood test items, the top eight features (partial Mayo score, MCH, BMI, BUN, concomitant use of AZA, lymphocyte fraction, height, and CRP) were selected for logistic regression to develop a prediction model for SFCR at week 22. In the validation using the external Cohort 2, the positive and negative predictive values of the prediction model were 54.5% and 92.3%, respectively. The prediction tool appeared useful for identifying patients with UC who would not achieve SFCR at week 22 during VDZ therapy. This study provides a proof-of-concept that machine learning using real-world data could permit personalized treatment for UC.

Download Full-text

Validity of social media for assessing treatment patterns in oncology patients: a case study in melanoma

JAMIA Open ◽

10.1093/jamiaopen/ooz013 ◽

2019 ◽

Vol 2 (4) ◽

pp. 416-422

Author(s):

Laura McDonald ◽

Varun Behl ◽

Vijayarakhavan Sundar ◽

Faisal Mehmud ◽

Bill Malcolm ◽

...

Keyword(s):

Social Media ◽

Confidence Intervals ◽

Real World ◽

Unmet Need ◽

Data Access ◽

Treatment Patterns ◽

Patient Treatment ◽

Real World Data ◽

The Real ◽

Third Line

Abstract There is a need to understand how patients are managed in the real world to better understand disease burden and unmet need. Traditional approaches to gather these data include the use of electronic medical record (EMR) or claims databases; however, in many cases data access policies prevent rapid insight gathering. Social media may provide a potential source of real-world data to assess treatment patterns, but the limitations and biases of doing so have not yet been evaluated. Here, we assessed whether patient treatment patterns extracted from publicly available patient forums compare to results from more traditional EMR and claims databases. We observed that the 95% confidence intervals of proportions of treatments received at first, second, and third line for advanced/metastatic melanoma generated from unstructured social media data overlapped with 95% confidence intervals from proportions obtained from 1 or more traditional EMR/Claims databases. Social media may offer a valid data option to understand treatment patterns in the real world.

Download Full-text

Establishment of Risk Prediction Model for Brain Metastasis with Breast Cancer by Using ctDNA Genetic Profile and Clinical Features from Real-World Data

SSRN Electronic Journal ◽

10.2139/ssrn.3403366 ◽

2019 ◽

Author(s):

Yu Tang ◽

Liping Lu ◽

Can Tian ◽

Ning Xie ◽

Jingyu Liu ◽

...

Keyword(s):

Breast Cancer ◽

Brain Metastasis ◽

Prediction Model ◽

Risk Prediction ◽

Clinical Features ◽

Real World ◽

Risk Prediction Model ◽

Genetic Profile ◽

Real World Data ◽

World Data

Download Full-text

Interpretable Machine Learning–Based Prediction of Intraoperative Cerebrospinal Fluid Leakage in Endoscopic Transsphenoidal Pituitary Surgery: A Pilot Study

Journal of Neurological Surgery Part B Skull Base ◽

10.1055/s-0041-1740621 ◽

2022 ◽

Author(s):

Pier Paolo Mattogno ◽

Valerio M. Caccavella ◽

Martina Giordano ◽

Quintino G. D'Alessandris ◽

Sabrina Chiloiro ◽

...

Keyword(s):

Machine Learning ◽

Cerebrospinal Fluid ◽

Logistic Regression ◽

Prediction Model ◽

Prediction Models ◽

Cerebrospinal Fluid Leakage ◽

Tumor Diameter ◽

Predictive Values ◽

Multivariable Logistic Regression ◽

Csf Leakage

Abstract Purpose Transsphenoidal surgery (TSS) for pituitary adenomas can be complicated by the occurrence of intraoperative cerebrospinal fluid (CSF) leakage (IOL). IOL significantly affects the course of surgery predisposing to the development of postoperative CSF leakage, a major source of morbidity and mortality in the postoperative period. The authors trained and internally validated the Random Forest (RF) prediction model to preoperatively identify patients at high risk for IOL. A locally interpretable model-agnostic explanations (LIME) algorithm is employed to elucidate the main drivers behind each machine learning (ML) model prediction. Methods The data of 210 patients who underwent TSS were collected; first, risk factors for IOL were identified via conventional statistical methods (multivariable logistic regression). Then, the authors trained, optimized, and audited a RF prediction model. Results IOL reported in 45 patients (21.5%). The recursive feature selection algorithm identified the following variables as the most significant determinants of IOL: Knosp's grade, sellar Hardy's grade, suprasellar Hardy's grade, tumor diameter (on X, Y, and Z axes), intercarotid distance, and secreting status (nonfunctioning and growth hormone [GH] secreting). Leveraging the predictive values of these variables, the RF prediction model achieved an area under the curve (AUC) of 0.83 (95% confidence interval [CI]: 0.78; 0.86), significantly outperforming the multivariable logistic regression model (AUC = 0.63). Conclusion A RF model that reliably identifies patients at risk for IOL was successfully trained and internally validated. ML-based prediction models can predict events that were previously judged nearly unpredictable; their deployment in clinical practice may result in improved patient care and reduced postoperative morbidity and healthcare costs.

Download Full-text

A machine learning-based prediction model of H3K27M mutations in brainstem gliomas using conventional MRI and clinical features

Radiotherapy and Oncology ◽

10.1016/j.radonc.2018.07.011 ◽

2019 ◽

Vol 130 ◽

pp. 172-179 ◽

Cited By ~ 6

Author(s):

Chang-cun Pan ◽

Jia Liu ◽

Jie Tang ◽

Xin Chen ◽

Fang Chen ◽

...

Keyword(s):

Machine Learning ◽

Prediction Model ◽

Clinical Features ◽

Conventional Mri

Download Full-text

Identifying patients at high risk for neutropenic complications during chemotherapy for metastatic breast cancer (MBC) with doxorubicin or pegylated liposomal doxorubicin: Development of a prediction model

Journal of Clinical Oncology ◽

10.1200/jco.2007.25.18_suppl.6598 ◽

2007 ◽

Vol 25 (18_suppl) ◽

pp. 6598-6598

Author(s):

G. Reardon ◽

D. Rayson ◽

J. Chang ◽

K. Gelmon ◽

G. Dranitsaris

Keyword(s):

High Risk ◽

Prediction Model ◽

Risk Prediction ◽

Metastatic Breast ◽

Pegylated Liposomal Doxorubicin ◽

Risk Scores ◽

Patient Treatment ◽

Poor Performance Status ◽

P Value ◽

Prediction Tool

6598 Background: Despite the effectiveness of anthracycline (ACH) therapy in the adjuvant and MBC settings, neutropenic complications (NC) remain a common and often unpredictable problem. Consequences may include dose reductions or delays in chemotherapy, or hospitalization for fever or infection. This study describes the development of a cycle-based risk prediction model for NC during chemotherapy with traditional doxorubicin (DOX) or a pegylated liposomal formulation (PLD) for MBC. Methods: Data analyzed was from a randomized clinical trial of MBC patients (n=509), who received chemotherapy with DOX (60 mg/m2 every 3 wks) or PLD (50 mg/m2 every 4 wks) [O'Brien, 2004]. NC were defined as an absolute neutrophil count (ANC) = 1.5 x106 cells/L, febrile neutropenia or neutropenia with infection. Patient, treatment and hematological factors potentially associated with NC were evaluated. Factors with a p-value of ≤ 0.25 within a cycle were included in a generalized estimating equations (GEE) regression model. Using backward elimination, we derived a risk scoring algorithm (range 0–63) from the final reduced model. Results: Risk factors retained in the model included poor performance status, ANC = 2.0 × 106 cells/L at some point in the previous cycle, the first cycle of chemotherapy, DOX vs. PLD and older age. A precycle risk score from = 25 to < 40 for a given patient was identified as being the optimal threshold for sensitivity (58.0%) and specificity (78.7%). Patients with a score at or beyond this threshold would be considered at high risk for developing NC in later cycles. Risk scores below, within, or above this threshold predict a 0.3%–2%, 3%–8% and a 9%–45% probability risk of NC, respectively. Conclusion: This risk prediction tool demonstrated acceptable internal validity and can be readily applied by the clinician prior to a given cycle of chemotherapy. The application of this prediction tool may allow for identification and targeted intervention (such as growth factor support or the use of PLD) for those most likely to experience NC during anthracycline-based chemotherapy for MBC. No significant financial relationships to disclose.

Download Full-text

P544 Comparison of the efficacy of a second intravenous or subcutaneous anti-TNF in the treatment of ulcerative colitis: Real-world data from the ENEIDA registry

Journal of Crohn s and Colitis ◽

10.1093/ecco-jcc/jjz203.672 ◽

2020 ◽

Vol 14 (Supplement_1) ◽

pp. S463-S464

Author(s):

P Torres-Rodriguez ◽

F Cañete ◽

M Calafat ◽

R Sánchez-Aldehuelo ◽

M Rivero ◽

...

Keyword(s):

Ulcerative Colitis ◽

Clinical Response ◽

Dose Escalation ◽

Biological Agents ◽

Lower Probability ◽

Clinical Activity ◽

Real World Data ◽

Loss Of Response ◽

Remission Rates ◽

Epidemiological Characteristics

Abstract Background Three anti-TNFs (one intravenous and two subcutaneous) are licensed for the treatment of ulcerative colitis (UC). However, it is not known if the efficacy of a second anti-TNF changes on whether it is intravenous or subcutaneous; this could justify the indication of biological agents with a different mechanism of action in second line. The aim of this study was to compare the efficacy of a second subcutaneous or intravenous anti-TNF in UC. Methods Patients from the prospectively maintained ENEIDA registry treated with consecutively intravenous and subcutaneous anti-TNF, who were naïve to biological agents, were identified. Patients were classified according to the administration route of the first anti-TNF in: IVi (intravenous initially) or SCi (subcutaneous initially). Patients treated for extraintestinal manifestations or pouchitis were excluded. Clinical activity and effectiveness were defined based on Partial Mayo Score (PMS) at baseline, 14 and 52 weeks. Loss of response, dose-escalation and treatment discontinuation were also assessed. Results 372 UC patients were included (270 IVi and 102 SCi). Both cohorts were similar in clinical-epidemiological characteristics, except for a higher proportion of patients with moderate-to-severe clinical activity at the beginning of the first anti-TNF in the IVi group (82% vs. 71%; p = 0.017) and at the beginning of the second anti-TNF (62% vs. 74%; p = 0.04). Clinical response and remission rates at week 14 for the second anti-TNF were 41% and 29% in IVi vs. 47% and 25% in SCi, respectively (p = ns). At week 52, clinical response/remission rates of the second anti-TNF were 37%/32% in IVi vs. 40%/29% in SCi (p = ns). A higher response rate at 14 weeks with the second anti-TNF was detected in the SCi group (40% vs. 68%; p = 0.012) when the reason for withdrawal of the first anti-TNF was secondary loss of response. The cumulative persistence of the second anti-TNF treatment in IVi and SCi were 55% and 54% after 1 year, and 41% and 40% after 2 years, respectively (p = ns). The SCi group had lower rates of dose-escalation with the second anti-TNF than IVi (34% and 29% in SCi vs. 57% and 49% in EVi, at 12 and 24 months, respectively -p = 0.004-). Dose-escalation of the first anti-TNF and moderate-to-severe clinical activity at the beginning of the second anti-TNF were associated with a lower probability of remission with the second anti-TNF in the short- and long-term. Conclusion The efficacy of infliximab after failure/intolerance of a subcutaneous anti-TNF is similar to that of subcutaneous anti-TNFs after infliximab failure/intolerance.

Download Full-text

Corrected ESR/Albumin Ratio as a Simple, Practical Marker Predicts the Severity in Egyptian Ulcerative Colitis Patients

The Open Biomarkers Journal ◽

10.2174/1875318302111010008 ◽

2021 ◽

Vol 11 (1) ◽

pp. 8-16

Author(s):

Shimaa El Sharawy ◽

Hemat El-Horany ◽

Ibrahim Amer

Keyword(s):

Ulcerative Colitis ◽

Sensitivity And Specificity ◽

Area Under The Curve ◽

High Sensitivity ◽

Clinical Activity ◽

Cross Sectional Survey ◽

Cross Sectional ◽

Predictive Values ◽

Albumin Ratio ◽

Egyptian Patients

Background: Serum biomarkers are commonly used for diagnosing and monitoring the disease activity of Ulcerative Colitis (UC) patients. However, their role in predicting disease severity among Egyptian patients is unknown. Objectives: The aim of this study was to correlate these biomarkers with clinical, endoscopic and histologic severity. Methods: This is a cross-sectional survey where 55 patients with UC were included to measure corrected Erythrocyte Sedimentation Rate (ESR), hematocrit (Hct), corrected ESR/albumin ratio and albumin, as well as colonoscopy and biopsy. Sensitivity and specificity, positive and negative predictive values were correlated with clinical, endoscopic, histologic severity. Results: The mean age of patients was 33 ± 8.4 years. In total, 27 (49.1%) were males and 28 (50.9%) were females. Area Under the Curve (AUC) values for the diagnosis of severe clinical disease were 0.947, 0.932, 0.727 and 0.685 for corrected ESR/albumin ratio, corrected ESR, Hct and albumin, respectively. Cut-off value to determine endoscopic severity for Hct was 34 (sensitivity: 88.89%, specificity: 83.78%, PPV: 72.7%, NPV: 93.9%, AUC: 0.963, p<0.001). Conclusion: Corrected ESR/albumin ratio was the best predictor of severe clinical activity of UC disease. Hct may be a marker of endoscopic and histological severity due to its high sensitivity and specificity as a diagnostic test.

Download Full-text

Deep Learning-Based Prediction Model for Breast Cancer Recurrence Using Adjuvant Breast Cancer Cohort in Tertiary Cancer Center Registry

Frontiers in Oncology ◽

10.3389/fonc.2021.596364 ◽

2021 ◽

Vol 11 ◽

Author(s):

Ji-Yeon Kim ◽

Yong Seok Lee ◽

Jonghan Yu ◽

Youngmin Park ◽

Se Kyung Lee ◽

...

Keyword(s):

Breast Cancer ◽

Machine Learning ◽

Deep Learning ◽

Prediction Model ◽

Real Time ◽

Prediction Models ◽

Unmet Need ◽

Absolute Error ◽

Data Set

Several prognosis prediction models have been developed for breast cancer (BC) patients with curative surgery, but there is still an unmet need to precisely determine BC prognosis for individual BC patients in real time. This is a retrospectively collected data analysis from adjuvant BC registry at Samsung Medical Center between January 2000 and December 2016. The initial data set contained 325 clinical data elements: baseline characteristics with demographics, clinical and pathologic information, and follow-up clinical information including laboratory and imaging data during surveillance. Weibull Time To Event Recurrent Neural Network (WTTE-RNN) by Martinsson was implemented for machine learning. We searched for the optimal window size as time-stamped inputs. To develop the prediction model, data from 13,117 patients were split into training (60%), validation (20%), and test (20%) sets. The median follow-up duration was 4.7 years and the median number of visits was 8.4. We identified 32 features related to BC recurrence and considered them in further analyses. Performance at a point of statistics was calculated using Harrell's C-index and area under the curve (AUC) at each 2-, 5-, and 7-year points. After 200 training epochs with a batch size of 100, the C-index reached 0.92 for the training data set and 0.89 for the validation and test data sets. The AUC values were 0.90 at 2-year point, 0.91 at 5-year point, and 0.91 at 7-year point. The deep learning-based final model outperformed three other machine learning-based models. In terms of pathologic characteristics, the median absolute error (MAE) and weighted mean absolute error (wMAE) showed great results of as little as 3.5%. This BC prognosis model to determine the probability of BC recurrence in real time was developed using information from the time of BC diagnosis and the follow-up period in RNN machine learning model.

Download Full-text