Specialization
Focus of research
Frank Bennis (1991) is an assistant professor in Clinical AI at the Emma Children’s Hospital of Amsterdam UMC, the Emma Center of Personalized Medicine and the Expertise Center for Gender Dysphoria. He was trained in Technical Medicine (BSc, MSc) at the University of Twente and obtained his PhD in 2020 from Maastricht University with a thesis on “Machine learning in Medicine”. Frank combines in-depth technical expertise with clinical knowledge, focusing on the development and implementation of advanced machine learning models for clinical prediction, with a particular emphasis on pediatrics and small sample sizes. Frank has experience supervising bachelor’s, master’s, and PhD students at the intersection of AI and healthcare. He served as coordinator and main lecturer for the master’s course “Machine learning for the quantified self” at Vrije Universiteit Amsterdam.
Frank develops clinical prediction models for a wide range of clinical problems, such as the prediction of bronchopulmonary dysplasia in preterm born infants, extubation success of children at the ICU and NICU, childhood neurocognitive and motor outcomes in children born preterm and probability of obesity development in patients with 16p11.2 gendeletion. He collaborates intensively with the Follow Me Emma Neuroscience Group (prof. dr. Jaap Oosterlaan and dr. Marsh Königs), the Emma Center of Personalised Medicine (prof. dr. Clara van Karnebeek and prof. dr. Mieke van Haelst), as well with the Expertise Center for Gender Dysphoria (prof. dr. Martin den Heijer). In addition, close collaboration exists with the neonatal and pediatric ICUs, the Data Science Expertise Centre and the Quantitative Data Analytics group at the Vrije Universiteit Amsterdam. Frank is also affiliated to the Follow Me health care innovation program of the Emma Children’s Hospital of Amsterdam UMC, an ambitious program that develops and implements care pathways for tertiary care patients with the ambition to improve care, support data-driven health care evaluation and innovation, facilitate clinical research to improve clinical care, and educate the future generation of health care professionals.
One of his main outcomes is the development and implementation of a bronchopulmonary dysplasia prediction model, which is currently as one of the first ML models in the AMC implemented and fully automated running within EPIC. The model has the potential to enable clinicians to directly visualize the risk of bronchopulmonary dysplasia the seventh day after birth, which is the most effective moment for treatment to start. Therefore, this enables clinicians to identify high-risk patients for treatment, whilst not giving unnecessary treatment with possible severe side-effect to low-risk patients. The model incorporates both clinical data available at birth with continuous vital-sign data during the first seven days in a deep learning model, which is unique in this field and leads to a high performance. This is currently in an early stage, not visible for clinicians, for which validation and development studies are being performed.
Valorisation
Frank actively contributes to the implementation of AI in clinical practice and the education of future healthcare professionals. His work results in directly applicable tools and insights for clinicians and supports the transition to data-driven and personalized medicine. Furthermore, he is active member of the junior board of the APH Digital Health program.