Towards sophisticated learning from EHRs: Increasing prediction specificity and accuracy using clinically meaningful risk criteria

Abstract

Abstract-Computer based analysis of Electronic Health Records (EHRs) has the potential to provide major novel insights of benefit both to specific individuals in the context of personalized medicine, as well as on the level of populationwide health care and policy. The present paper introduces a novel algorithm that uses machine learning for the discovery of longitudinal patterns in the diagnoses of diseases. Two key technical novelties are introduced: one in the form of a novel learning paradigm which enables greater learning specificity, and another in the form of a risk driven identification of confounding diagnoses. We present a series of experiments which demonstrate the effectiveness of the proposed techniques, and which reveal novel insights regarding the most promising future research directions.

Identification of promising research directions using machine learning aided medical literature analysis

“…Large amounts of highly heterogeneous data types are pervasive in medicine. Usually the concept of so-called "big data" in medicine is associated with the analysis of Electronic Health Records [14], [4], [7], [22], [23], large scale sociodemographic surveys of death causes [19], social media mining for health related data [12] etc. Much less discussed and yet arguably no less important realm where the amount of information presents a challenge to the medical field is the medical literature corpus itself.…”

Towards objective and reproducible study of patient-doctor interaction: Automatic text analysis based VR-CoDES annotation of consultation transcripts

“…For the SVM classifier we used the Gaussian kernel with the scale set to 0. 25 √ n p where n p is the number of predictors. …”

“…In particular, we address the problem of labelling a patient's utterance (verbal communication, uninterrupted by the HCP, which comprises a single or multiple sentences) as containing (i) an explicit concern, (ii) an emotional cue, or (iii) neither. Thus, in addition to addressing the specific issues of patient-HCP communication, the present work contributes to the overall effort in the use of modern computer science in assisting health care provision [4], [7], [24], [25].…”

Light Curve Analysis From Kepler Spacecraft Collected Data

“…Indeed, a great and increasing amount of science now relies on the analysis of large quantities of data [2,6,35]. Signi cant e orts in the realm of personalized medicine, for example in the analysis of large scale electronic health records [3,30,37] have already demonstrated highly promising results.. e highly multi-modal nature of such data [5] which may consist of 'conventional' or infrared images, depth information, physical measurements of di erent types, demographic information, and numerous other forms, as well as the domain speci c semantic gap interlaced with the interpretation of the aforementioned information, all also present major research challenges. Notwithstanding the breadth of e orts touched upon above, there are many scienti c areas in which the use of state of the art arti cial intelligence remains li le explored, arguably in no small part because they are (o en incorrectly) seen as having limited practical relevance.…”