ML & AI, image classification and recognition for the remote take charge of patients and the improving of Melanoma early diagnosis
Artificial Intelligence and Big Data change the patient management model to which a path of prevention and early diagnosis of Melanoma is applied. The goal is to focus on the patient, improve the accuracy in the prevention and diagnosis of the territorial hospitals, break down the waiting lists, up to "socialize" and make efficient the services of taking charge by modifying the interaction between Hospital-Territory-Patient.
The focus is on "precision" prevention campaigns with the Real World Evidence of Big Data for Preventive, Predictive, Personalized and Participatory digital healthcare.Trained deep neural networks and Deep Learning techniques obtain more accurate precision than dermatologists on Melanoma with the analysis of dermoscopic images and patient's information (age, sex, phototype, residence, familiarity, lifestyles).Trained algorithms improve the diagnostic accuracy and support the doctors.
The project therefore aims the creation of a Digital Platform for the remote take charge of patients, the collection of data and images from heterogeneous sources (BD), their processing using AI/DL algorithms to support non-expert doctors and the Hub center.
The patient can be taken in charge and enrolled in different innovative ways through:
-sending images from a smartphone with a mobile App (which sets the standard criteria for image acquisition and collects information for anamnesis);
-acquisition of images with automatic videodermatoscopy (self-service) systems by expanding the Dermatoscopy, a Melanoma diagnostic tool, from the Hub to territory.
The data collected on the Platform feed the databases to train the algorithms used to select the patients to be visited and to define the target of the prevention campaign.The patient's report of a suspect neo, validated by algorithm, leads to taking charge directly in the Hub, determining timely treatment and care.
The first results of the Project consist in the realization of a proof of concept and its experimentation in the Calabria Region, because the less advanced systems as in Southern Italy have a negative spread of 43.5%, compared to those of the North in the incidence of Melanoma (Airtum 2018). This difference is determined by a greater capacity to prevent the health systems of the Center-North.
The proposed system costs much less than the recruitment and training of specialists not present on the market today and allows to reduce the incidence of the disease, treatment costs, mortality and social costs.
2019, Demetrio Naccari Carlizzi et al., ML & AI, image classification and recognition for the remote take charge of patients and the improving of Melanoma early diagnosis, in Proceedings Convegno Internazionale 2019 3rd ICEHTMC – International Clinical Engineering and Health Technology Management Congress. pre-print. http://www.icehtmc.com/