There is a growing interest in representing patient histories as data rather than just text. In the UK, for example, this includes the NHS Spine that is part of Connecting for Health (a national initiative intended to facilitate clinical management), and the use for research purposes of large patient databases such as the General Practice Research Database (GRPD) [1] and The Health Improvement
Network (THIN) [2]. The GRPD alone contains coded diagnostic, demographic and prescribing information for over 12.5 million patients from around 620 practices around selleck chemical the UK (approx. 7% of all UK practices). It has not gone unnoticed that one of the many advantages of representing clinical histories in this way is the ability to perform in silica experiments through statistical studies on data aggregated across patient populations. We explore here the possibilities for exploiting this data for yet another purpose: producing textual summaries of the history of individual patients automatically from the data. A facility such as this could mean that rather than having to wade through
BLZ945 research buy collections of paper documents that make up a typical “patient record”, or grapple with a complex database, practitioners would have at their disposal a new form of Electronic Patient Record that provides a customised view of a patient’s history. Current studies show that the quality of healthcare outcomes increases when doctors are able to spend sufficient time with patients to explore their symptoms, explain their condition and negotiate their treatment plan [3]. There is also agreement that the available duration of consultations in the UK is typically only between 7 and 10 min [4]. In most UK practices, clinicians are allocated 10 min for existing patients and 20 min for new patients. Since a consultation session also includes the time spent by the clinician familiarising
herself with the patient’s condition, the more time that is devoted to that activity, the less there will be for interacting with the patient. Since clinicians have only a limited Glutamate dehydrogenase time for each patient, and since they clearly cannot be expected to be database experts, their ability to receive the information they require in an easily digestible form is obviously critical. A popular approach to this problem has been to build systems that produce graphical visualisations of the underlying patient data [5], [6], [7], [8], [9] and [10], but recent studies have shown that graphical visualisations of medical data are not always helpful for clinicians’ decision-making [11] and [12]. Instead, we have chosen to explore the use of textual summaries, relying on the familiarity of this medium for presenting medical records.