EP171 - Artificial intelligence-driven high-resolution anorectal manometry – data from atransatlantic study

ePoster Presenter(s)

• MM

  Miguel Mascarenhas, n/a

  Gastroenterology, Centro Hospitalar Universitario de Sao Joao
  Hospital São João, Porto.
  Porto, Porto, Portugal

  All disclosed relationships have been reviewed and managed prior to the start of this activity. Any relevant relationships will be listed below. This individual has no financial relationships to disclose

  View ePoster

Author(s)

• AC

  António Costa, MD

  Department of Gastroenterology, Puerta del Hierro University Hospital
  .
  ., Madrid, Spain

  Disclosure information not submitted.
• MC

  Matheus Carvalho, MD

  Dr
  Hospital das Clinicas da Faculdade de Medicina da Universidade de São Paulo de Ribeirão Preto
  São Paulo, Sao Paulo, Brazil

  Disclosure information not submitted.
• JM

  Joana Mota, n/a

  Dr
  Hospital São João, Porto.
  Porto, Porto, Portugal

  Disclosure information not submitted.
• FM

  Francisco Mendes, .

  Department of Gastroenterology, São João University Hospital
  .
  ., Porto, Portugal

  Disclosure information not submitted.
• MM

  Miguel Martins, n/a

  Engineer
  Faculty of Engineering of the University of Porto
  Porto, Porto, Portugal

  Disclosure information not submitted.
• PC

  Pedro Cardoso, MD

  Dr
  Centro Hospitalar Universitário de São João, Gastroenterology, Porto, Portugal
  Porto, Porto, Portugal

  Disclosure information not submitted.
• JC

  João Cordeiro, MD

  Attending
  Department of Gastroenterology, Precision Medicine Unit, Centro Hospitalar Universitário São João,
  Lisboa, Lisboa, Portugal

  Disclosure information not submitted.
• IM

  Ismael Martinez, MD

  Attending
  Hospital Universitario Puerta de Hierro Majadahonda,
  Madrid, Madrid, Spain

  Disclosure information not submitted.
• BN

  Benjamin Niland, MD

  Attending
  Division of Gastroenterology, University of South Alabama
  Alabama, Alabama, United States

  Disclosure information not submitted.
• CS

  Cecilio Santander, MD

  Attending
  Hospital Universitario La Princesa
  Madri, Madrid, Spain

  Disclosure information not submitted.
• VR

  Virginia Royo, MD

  Attending
  Hospital Universitario Puerta de Hierro Majadahonda
  Madrid, Madrid, Spain

  Disclosure information not submitted.
• Jd

  Jack di Palma, MD

  Attending
  Division of Gastroenterology, University of South Alabama,
  Alabama, Alabama, United States

  Disclosure information not submitted.
• GM

  Guilherme Macedo, .

  Department of Gastroenterology, São João University Hospital
  .
  ., Porto, Portugal

  Disclosure information not submitted.

Purpose/Background: Anorectal functional disorders (AFD) are common in clinical practice, with high-resolution anorectal manometry (HR-ARM) being the gold standard for their evaluation. Yet, HR-ARM is limited by its restricted accessibility and complex data analysis. This study aims to develop and validate an artificial intelligence (AI) model for the identification, differentiation of disorders of tone and contractility in HR-ARM.

Methods

960 HR-ARM exams from 2 device manufacturers and 3 centers in Spain and USA were used for the development and testing of multiple machine learning (ML) algorithms. AFD were classified according to the London Classification. Exams were divided into training and testing datasets with an 80%/20% patient split design. The models' prediction was compared to a consensus-based classification by two HR-ARM experts. The testing dataset was used to evaluate model performance through its accuracy and the area under the receiver-operating characteristic curve (AUC-ROC). Results The Gradient Boosting Classifier (GBC) showed the best performance, identifying disorders of anal tone and contractility with 85.6% accuracy and an AUC-ROC of 0.91. LGBMClassifier detected anal normotension with hypocontractility with 90.5% accuracy. AdaBoostClassifier identified anal hypotension with hypocontractility with 94.9% accuracy. Anal hypotension with normal contractility was detected by the LGBMClassifier with 94.1% accuracy. Anal hypertension was identified by the GBC with 96.9% accuracy.

Conclusion ML models accurately identified and differentiated disorders of anal tone and contractility in HR-ARM. Data inclusion from 2 device manufacturers addresses the challenge of interoperability, while including patients from 2 continents mitigates demographic bias. AI-driven HR-ARM has the potential to optimize exam accessibility and accuracy, enhancing patients's management with AFD.

Methods/Interventions: 960 HR-ARM exams from 2 device manufacturers and 3 centers in Spain and the United States of America were used for the development and testing of multiple machine learning (ML) algorithms. Anorectal motility disorders were classified according to the London Classification. Exams were divided into training and testing datasets with an 80%/20% patient split design. The models' prediction was compared to a consensus-based classification by two HR-ARM experts. The testing dataset was used to evaluate model performance through its accuracy and the area under the receiver-operating characteristic curve (AUC-ROC).

Results/Outcomes: The GradientBoostingClassifier showed the best performance, identifying disorders of

anal tone and contractility with 85.6% accuracy and an AUC-ROC of 0.91. The LGBMClassifier detected anal normotension with hypocontractility with 90.5% accuracy. AdaBoostClassifier identified anal hypotension with hypocontractility with 94.9% accuracy. Anal hypotension with normal contractility was detected by the

LGBMClassifier with 94.1% accuracy. Anal hypertension was identified by the

GradientBoostClassifier with 96.9% accuracy.

Conclusion/Discussion: ML models accurately identified and differentiated disorders of anal tone and

contractility in HR-ARM. The inclusion of data from two device manufacturers addresses the challenge of interoperability, while including patients from two continents mitigates the demographic bias. AI-driven HR-ARM has the potential to optimize exam accessibility and accuracy, enhancing the management of patients with functional anorectal disorders.

Figure 1
Evaluation of the resting pressure and squeeze maneuver, with examples of a normal exam (A), anal hypotension with normal contractility (B), anal hypertension (C), anal normotension with hypocontractility (D) and combined anal hypotension and hypocontractility (E).