The AIDPRO-CRC Trial

Introduction:

The AIDPRO-CRC trial is an investigator-initiated nationwide multicenter randomized controlled trial. The trial aims to investigate the clinical effects of an AI-augmented solution "AIDPRO manual CRC" for optimization of perioperative treatment by personalized risk stratification of patients undergoing CRC surgery. The protocol adheres to the SPIRIT Statement recommendations.

The AIDPRO-CRC trial is a pre-market, pivotal stage, confirmatory clinical investigation designed to evaluate the safety and effectiveness of the AIDPRO manual CRC algorithm. As a pivotal clinical investigation, this study is critical for generating the robust evidence required to support regulatory submissions for CE marking. The investigation involves an interventional approach, meaning that participants will undergo specific procedures or treatments as part of the study. This design has been specifically chosen to rigorously assess the performance of the AIDPRO manual CRC device in a real-world clinical setting, providing the necessary data to demonstrate its safety and effectiveness. The results from this trial will be used to seek CE marking, enabling the AIDPRO manual CRC to be brought to market in the future.

Objective of the Study:

Colorectal cancer (CRC) is the second leading cause of cancer-related mortality worldwide. Despite advances in standardized treatment protocols, significant challenges remain in reducing complications, readmissions, and mortality. Addressing these challenges necessitates a transition toward individualized, data-driven treatment strategies.

The AIDPRO-CRC trial aims to evaluate the effectiveness of using artificial intelligence (AI) to distinguish between high- and low-risk patients to offer a personalized and optimized perioperative care pathway for individuals undergoing surgery for colon and rectal cancer. This pathway is tailored based on each patient's specific risk factors and individual clinical profile.

The trial compares risk assessments made by a surgeon unaided to those made by a surgeon supported by an AI model.

For surgeons, achieving a comprehensive understanding of the numerous factors influencing a patient's postoperative risk can be both complex and time-consuming. This study therefore investigates whether AI-assisted risk assessment can improve the allocation of healthcare resources while simultaneously optimizing patient outcomes.

We aim to evaluate the following three primary hypotheses:

* AI-based decision support will enable a more efficient allocation of healthcare resources, potentially improving the cost-effectiveness of care delivery. * Surgeons and other healthcare professionals will find the AI-enabled software acceptable and user-friendly, and will perceive it as adding value compared to conventional decision-making processes. * Surgeons using AI assistance will recommend more consistent treatment pathways than when relying solely on their usual clinical judgment.

Additionally, the study aims to assess the impact of AI-based risk stratification on the incidence of complicated postoperative courses in cancer patients.

The concept of AI-assisted risk assessment for colorectal cancer patients - enabling individualized perioperative optimization based on risk profiles - has already been pilot-tested at the Department of Surgery, Zealand University Hospital, Køge, under the name AID-SURG. This pilot has been fully implemented for approximately two years. Preliminary results suggest improvements in patient outcomes, including reductions in complications, readmissions, and length of hospital stay. A scientific manuscript describing these findings is currently under peer review for publication in an international journal.

Study Methodology, Design, and Procedures:

The AIDPRO-CRC trial is a nationwide, multicenter randomized clinical trial conducted across hospitals in all five Danish regions. The primary objective is to evaluate whether an artificial intelligence (AI)-driven solution can improve the treatment of patients undergoing surgery for colorectal cancer by tailoring perioperative care to the individual patient's risk profile. In addition to the randomized trial, the study comprises a questionnaire-based survey among healthcare professionals and a simulated substudy. The study has several aims:

* To determine the cost and clinical impact of using a digital decision-support platform powered by AI for risk assessment in surgical oncology. (Randomized controlled trial with patients) * To explore how clinicians and other healthcare professionals perceive the AI tool in terms of usability and relevance in clinical practice. (Questionnaire survey) * To examine whether treatment recommendations differ depending on whether clinicians use the AI-enabled decision-support platform. (Simulated substudy) Seven hospitals across Denmark are participating in the patient inclusion phase, while additional healthcare staff at these sites may participate in the questionnaire and simulated substudy. One hospital acts as the coordinating center but does not include patients.

Randomized Controlled Trial (RCT) with Patients:

All patients with suspected or confirmed colon and/or rectal cancer at the participating centers will be screened for eligibility. Only patients with a confirmed diagnosis and an indication for curative-intent surgery are eligible for inclusion. All patients will continue to receive care in accordance with national clinical guidelines and the colorectal cancer fast-track program.

Participants will be randomized to one of two risk-assessment arms:

* An AI-supported risk assessment arm * A standard surgeon-assessed risk assessment arm The goal is to optimize perioperative treatment (pre-, intra-, and postoperative) to reduce postoperative complications. The intensity of the optimization offered is scaled based on the assigned risk level, whether determined by the surgeon or by the AI model. Patients are allocated to one of four predefined perioperative care packages (A, B, C, or D), which are already part of standard care across participating sites. Therefore, patients who decline participation will still receive personalized care based on these four packages.

The only difference between the study arms is the method of risk assessment used to guide the treatment plan: either an AI-based model or the surgeon's clinical judgment. All other treatment components follow national guidelines and are identical across all sites.

The aim of the intervention is to determine whether AI-assisted risk stratification can improve the allocation of patients into the four risk-based treatment groups, ensuring that only patients with a high risk of complications receive more intensive optimization, thus improving the efficiency of healthcare resource use.

Study Arms in the RCT AI-based Risk Assessment (Intervention Group)

* Description: An advanced AI model functions as a decision-support tool to estimate each patient's perioperative risk. * Purpose: The AI model uses various patient-specific data inputs to predict risk and assign a tailored care pathway, based on a large historical dataset. * Expected Outcome: The use of AI is expected to improve the precision of risk stratification, thereby optimizing resource utilization.

Surgeon-based Risk Assessment (Control Group)

* Description: Experienced colorectal surgeons assess patient risk based on clinical judgment and national guidelines. * Purpose: Traditional clinical assessment is used to assign patients to the appropriate care pathway. * Expected Outcome: This arm serves as the clinical standard-of-care comparator against which the effectiveness of AI-guided decision-making is evaluated.

Perioperative Care Packages (A-D)

These four standardized treatment packages are part of routine clinical practice and are not unique to the AIDPRO-CRC trial. The package assigned depends on the risk assessment method and stratifies patients as follows:

* Package A: Low-risk patients * Package B: Moderate-risk patients * Package C: High-risk patients * Package D: Very high-risk patients

Each package includes:

* Preoperative optimization: Multidisciplinary interventions (e.g., surgeons, anesthesiologists, physiotherapists, geriatricians, and dieticians) to enhance patients' physical readiness for surgery. * Intraoperative care: Standardized surgical techniques and anesthetic protocols designed to minimize physiological stress and reduce complications. * Postoperative care: Enhanced recovery protocols emphasizing early mobilization, pain management, and nutritional support to promote faster recovery.

Healthcare Professional Involvement - Usability and Simulation Studies:

To investigate the final two hypotheses, healthcare professionals involved in using the AI model during the study period will participate in both a usability survey and a simulated decision-making substudy.

Usability Survey Two months after patient inclusion begins at each site, all relevant users will receive an email invitation to complete a questionnaire regarding their experience with the MDTPlatform. Prior to participation, informed consent and eligibility confirmation are required. The survey is repeated after one year and again at study completion.

Simulated Substudy Eligible physicians are contacted by a designated researcher and receive information about the simulated decision-making exercise. Upon consent, participants complete a simulation involving 20 patient cases: 10 cases in a standard format (e.g., Word documents) and 10 via the MDTPlatform with AI support. Each case is presented only once to each physician-either in the standard format or via the platform. Physicians indicate their treatment recommendation for each case. If necessary, multiple sessions may be arranged. Physicians may assess multiple cases until all 75 cases have been reviewed by at least one participant.

Patients

All patients referred to a surgical department with a confirmed first-time diagnosis of colon and/or rectal cancer and deemed eligible for potentially curative surgery will be offered participation in the study. Participants will be randomly assigned to one of two groups:

1. Intervention group: Risk assessment is performed using an artificial intelligence (AI)-based prediction model to estimate the risk of 1-year postoperative mortality. 2. Control group: Risk assessment is performed by an experienced colorectal surgeon based on clinical judgment and standard risk evaluation practices.

Based on the estimated risk, patients are assigned to one of four predefined perioperative care pathways (Packages A-D). Patients who decline consent for risk estimation will not be enrolled in the trial but will still be offered optimized perioperative care as per local standard clinical protocols.

Sample size estimation - patients Sample size was calculated based on data from the AID-SURG pilot project to detect a cost-saving effect of the AI-based risk stratification. A simulation-based power analysis determined that 600 participants per group (1,200 total) are needed to achieve 93% statistical power with a 5% significance level, to detect a cost reduction of $94.9 USD per patient in the AI group compared to the control group, assuming a tendency of surgeons to allocate more patients to high-risk groups.

No research biobank will be established, and no biological samples will be collected or used from existing biobanks in this study.

• The rate of complicated postoperative course 90 days after surgery [Time frame: 90 days post-operative]
• Postoperative Complications [Time frame: 90 days Post-Operative]
• Length of Hospital stay (LOS) > 4 days [Time frame: 90 days post-operative]
• Readmission Rates [Time frame: 90 days postoperatively]
• Days Alive and Out of Hospital 30 days and 90 days [Time frame: 90 Days Postoperatively]
• Composite outcomes: [Time frame: 90 days post-operative]
• Time from MDT to Surgery [Time frame: Preoperative]