Thera-Centre for Language and Computation, a University of Barcelona spin-off located at the Barcelona Science Park, announces the development of a software program that can interpret the natural written language used by doctors and other health professionals and automatically encode it. The innovative tool, already in use at the Puigvert Foundation of Barcelona, has a level of reliability that tops 95%. Not only does it help to improve information management and the control exercised by health centres over their own activity, but it also provides a more reliable record on the incidence of diseases and expands knowledge relating to their symptomatology and evolution.

Thera’s encoding system, known as hCOD, is based on artificial intelligence and language processing technologies. The system’s level of resolution enables it to detect information contained in diagnoses that cannot be encoded because of lack of clinical information. It can also deal with multiple diagnoses, correct misspellings and even work simultaneously in multiple languages. The system is already working with the international standard ICB 9 CM (The International Classification of Diseases, 9th Revision, Clinical Modification) “), which was approved by the World Health Organisation (WHO). In addition, it has been designed to adapt to various internationally validated encoding standards, such as the one used by nursing personnel (NANDA), or SNOMED CT, devised by the International Healthcare Terminology Standards Development Organisation (IHTSIDO). SNOMED CT, for example, includes over a million entries and 300,000 codes, enabling it to encode diagnoses and medical information with greater precision

To date, automated encoding programs have generally been based on systems of terminological coincidence. In the actual environment found in health centres, however, doctors use language freely and that puts limits on the likelihood of correctly encoding all the information involved. Moreover, the terminology used in the standard systems and any writing errors that may occur—i.e., misspellings and typographical mistakes—are very wide elements of dispersion, hindering the development of an automated system based solely on technology using terminological coincidence.

As the tool can interpret the expressions and abbreviations regularly used by health professionals and also adapt to any system of encoding, it can be used in all areas of medical care, including primary and clinical care, x-ray services, emergency services and out-patient visits. The encoded information can then be put into a central database and subjected to query, easing the work of documentation departments and increasing the volume of information and knowledge available on a centre’s activity. As a result, the service can be improved and resources saved. “Our aim in creating the system has been to devise a tool that not only makes the job of encoding and managing the information more effective but also gives something back to the medical team. We want them to see the effort put into encoding pay off in more useful information, improvements to healthcare services and better research carried out in the centre”, explains Enric Castelló, director of Thera.