A scientific research carried out by Consorzio Arsenàl.IT in collaboration with Noovle was presented in Milan last May at the Google Cloud Summit. The project tested the use of Machine Learning algorithms to automatically classify digital clinical documents and extract the largest possible number of significant clinical information from unstructured ones.
Arsenàl.IT is Centro Veneto Research and Innovation for Digital Health, which, since 2012, on behalf of Regione Veneto and Azienda Zero, coordinates the regional electronic health file (FSEr) project, guaranteeing in particular all the technical aspects related to interoperability of health information systems. One of the reasons that led to the partnership with Noovle and Google is that for over a decade, Arsenàl.IT has been collaborating with international organizations to develop the infrastructure of the FSEr, ensuring its adherence to the most advanced standards and, at the same time, the possibility of feeding in real time an infrastructure for secondary data use (big data) for scientific research purposes.
In this context, with the implementation of the FSEr, a series of information has been made available in the documents both in a structured format (such as the laboratory report drawn up according to the CDA® standard) and unstructured (for example, hospital discharge letters in PDF format).
The research envisages the use of Machine Learning techniques for the automatic identification of the salient parts of the discharge letters in order to automatically identify the diagnoses present in each section according to the international ICD-9-CM coding.
The objective is twofold. One is to understand how much and how the algorithms of Machine Learning and Artificial Intelligence solutions developed in other fields are applicable to the clinical health one. The other is to develop specific Machine Learning algorithms to be transformed into services connected to the FSEr able to offer an added value both to the clinical field (developing support applications for clinical decisions) and to regional governance (population medicine, prevention, etc.). The importance and the potential that the results of the research can have from predictive and preventive medicine, which offers advantages to the entire population, are easily understood.
There are many fundamental elements that guarantee safety in the use of clinical-health data in this research project. In particular, it should be noted that the design of the technological infrastructure provides the distinction between the world of the dossier (in which the data are clear and used for clinical purposes) and the world of big data, in which the data , after being carefully anonymized , are inserted and used for research purposes . All this as required by national legislation on the subject.
The results of the project, which will be completed by the end of this year, will be announced after the careful analysis of the application of the algorithm being studied in over 70,000 letters of hospital discharge.
Success stories
Arsenàl.IT
The project launched with Noovle plan to use Machine Learning techniques for the automatic identification of the salient parts of the discharge letters in order to automatically identify diagnoses present in each section according to the international classification ICD-9-CM.
Noovle, the first Google Cloud Premier Partner in Europe, invites Maltese stakeholders to an exclusive summit dedicated to Google Cloud technologies on Friday, 19 October, at the Castello dei Baroni, Wardija.
Noovle, a Premier Google Cloud Partner, is an ICT company that specialises in web services and has considerable expertise in Digital Transformation and Integration System projects. It also manages the BNoovle Cloud Academy.
For further information, please contact Margaret Brincat at [email protected] or call 9940 6743