Colorectal Cancer
Conditions
Keywords
Young-onset colorectal cancer, Cancer screening, Artificial intelligence, Machine learning
Brief summary
In this study, we aimed to develop, internally and temporally validate the machine learning models to help screen YOCRC bansed on the retrospective extracted Electronic Medical Records (EMR) data.
Detailed description
Diagnosis of young-onset colorectal cancer (YOCRC) has become more common in recent decades. Screening CRC among younger adults still remains a challenge. In this study, We plan to retrospectively extracte the relevant clinical data of young individuals who underwent colonoscopy from 2013 to 2022 using Electronic Medical Record (EMR). Multiple supervised machine learning techniques will be applied to distinguish YOCRC and non-YOCRC individuals, the above classifiers will be trained and internally validated in the training dataset and internal validation dataset admitted between 2013 and 2021, respectively. We will also assess the temporal external validity of the classifiers based on the admissions from 2022.
Interventions
DIAGNOSTIC_TESTUsing routine clinical data and machine learning models.
This study used clinical data and machine learning model to screen young-onset colorectal cancer.
Sponsors
Renmin Hospital of Wuhan University
Study design
Observational model
OTHER
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to 49 Years
Healthy volunteers
Yes
Inclusion criteria
* Newly diagnosed with CRC (YOCRC group) * Age at 18-49 when diagnosis (YOCRC group) * Never received any CRC-related treatment (YOCRC group) * No CRC confirmed by colonoscopy or pathology (non-YOCRC group) * Age at 18-49 (non-YOCRC group)
Exclusion criteria
* Hospital stay less than 24 hours or with incomplete Complete Blood Count * Patients with inflammatory bowel disease or hereditary CRC syndromes * History of other types of primary malignant tumor and other reasons that made them unsuitable for enrollment
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| The performance of machine learning screening models | through study completion, an average of 1 year | The performance of young-onset colorectal cancer screening models will be assessed by calculating the area under the receiver operating characteristic (ROC) curve (AUC), Accuracy, Recall, Specificity, Negative predictive value (NPV), Positive predictive value (PPV, or called Precision). |
Countries
China
Outcome results
None listed