Most diagnosis and medical decision-making processes for the purpose of treatments are commonly based on the recorded medical sentiments from an earlier treatment stage. These medical sentiments are in turn based on patient (subjective) narration and treatment results (medical letters, examination reports, radiological reports, and so on). However, the electronic medical record (EMR) is becoming an important tool in the medical treatment process by helping with quick medical decision making for further diagnosis and treatment. Unfortunately, most existing EMR systems do not provide any feature to help the medical decision maker(s) analyze automatically, or autonomously get, a medical sentiment.

Medical sentiments are more complex than general text. Words, sentences, and expressions used to write medical sentiments need to be thoroughly chosen and have to provide an accurate and less subjective insight into the patient’s medical status in order to ease further treatment.

The authors of this paper perform, contrary to and beyond the existing sentiment analysis methodologies/processes in clinical or medical contexts, a quantitative analysis of sentiments in clinical context by analyzing the language and sentiment expressions in clinical narratives and medical social media. Actually, existing sentiment analyses are performed on medical web contents, biomedical literature, and clinical notes. The authors provide through this work new opportunities and approaches in sentiment analysis in the medical context by considering more medical reports, such as nurse letters, radiological reports, discharge summaries, and drug reviews, as well as user-generated texts such as blogs and interviews (for example medblogs and Slashdot interviews).

Through this research work, the authors greatly contribute to expanding the methodology of analyzing sentiments in the medical context. The authors have performed a quantitative comparison of word usage, which shows that the language used in medical documents is more objective than that used in medical social media. This paper could potentially contribute to improving the existing EMR. I therefore recommend it to scientists working in the fields of e-health, health informatics, and computer linguistics.