Carlo M Bertoncelli1,2, Paola Altamura3, Edgar Ramos Vieira4, Domenico Bertoncelli5, Michal Latalski6, Stéphanie Berthet7, Federico Solla1. 1. Department of Pediatric Surgery, Lenval University Children's Hospital of Nice, Nice, France. 2. EEAP H Germain, Fondation Lenval - Children Hospital, Nice, France. 3. Department of Medicinal Chemistry and Pharmaceutical Technology, University of Chieti, Chieti, Italy. 4. Department of Physical Therapy, Florida International University, Miami, Florida, USA. 5. Department of Information Engineering, Computer Science and Mathematics, University of L'Aquila, L'Aquila, Italy. 6. Children Orthopedic Department, Children University Hospital of Lublin, Lublin, Poland. 7. Pediatric Gastroenterology Unit, Lenval University Children's Hospital of Nice, Nice, France.
Abstract
BACKGROUND: Factors associated with gastrostomy placement in adolescents with developmental disabilities (DDs) and cerebral palsy (CP) are poorly investigated. We aimed to develop and validate a machine learning (ML) model for gastrostomy placement in adolescents with DDs and CP. METHODS: We performed a multinational, double-blinded, case-control study including 130 adolescents with severe DD and CP (72 males, 58 females; mean age 16 ± 2 years). Data on etiology, diagnosis, spasticity, epilepsy, clinical history, and functional assessments such as the Eating and Drinking Ability Classification System, Manual Ability Classification System, and Gross Motor Function Classification System were collected between 2005 and 2015. Analysis included Fisher exact test, multiple logistic regressions, and a supervised ML model, named PredictMed, to identify factors associated with gastrostomy placement. "Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis" guidelines were followed. RESULTS: Poor motor function (P < 0.001), trunk muscle tone disorder (P < 0.001), male gender (P < 0.01), epilepsy (P = 0.01), and severe neuromuscular scoliosis (P = 0.04) were factors linked with gastrostomy placement in univariate analysis. Epilepsy (P = 0.03), poor motor function (P = 0.04), and male gender (P = 0.04) were associated with gastrostomy placement in multivariate analysis with 95% accuracy. CONCLUSION: Epilepsy, poor motor function, trunk muscles tone disorder, and male gender were accurate, sensitive, and specific factors associated with gastrostomy need.
BACKGROUND: Factors associated with gastrostomy placement in adolescents with developmental disabilities (DDs) and cerebral palsy (CP) are poorly investigated. We aimed to develop and validate a machine learning (ML) model for gastrostomy placement in adolescents with DDs and CP. METHODS: We performed a multinational, double-blinded, case-control study including 130 adolescents with severe DD and CP (72 males, 58 females; mean age 16 ± 2 years). Data on etiology, diagnosis, spasticity, epilepsy, clinical history, and functional assessments such as the Eating and Drinking Ability Classification System, Manual Ability Classification System, and Gross Motor Function Classification System were collected between 2005 and 2015. Analysis included Fisher exact test, multiple logistic regressions, and a supervised ML model, named PredictMed, to identify factors associated with gastrostomy placement. "Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis" guidelines were followed. RESULTS: Poor motor function (P < 0.001), trunk muscle tone disorder (P < 0.001), male gender (P < 0.01), epilepsy (P = 0.01), and severe neuromuscular scoliosis (P = 0.04) were factors linked with gastrostomy placement in univariate analysis. Epilepsy (P = 0.03), poor motor function (P = 0.04), and male gender (P = 0.04) were associated with gastrostomy placement in multivariate analysis with 95% accuracy. CONCLUSION:Epilepsy, poor motor function, trunk muscles tone disorder, and male gender were accurate, sensitive, and specific factors associated with gastrostomy need.
Authors: Federico Solla; Walid Lakhal; Christian Morin; Jerome Sales de Gauzy; Gaby Kreichati; Ibrahim Obeid; Stéphane Wolff; Joël Lechevallier; Henry F Parent; Jean-Luc Clément; Carlo M Bertoncelli Journal: Eur J Orthop Surg Traumatol Date: 2021-06-18