Search

A Chatbot or Conversational Agent is a computer application that simulates the conversation with a human person (by text or voice), giving automated responses to people's needs. In the healthcare domain, chatbots can be beneficial to help patients, as a complement to care by health personnel, especially in times of high demand or constrained resources such as the COVID-19 Pandemic. In this paper we share the design and implementation of a healthcare chatbot called Tana at the Hospital Italiano de Buenos Aires. Considering best practices and being aware of possible unintended consequences, we must take advantage of information and communication technologies, such as chatbots, to analyze and promote useful conversations for the health of all people.

WHO and UNICEF highlight vaccination as the most cost-effective method of prevention of infectious diseases. An effective public health strategy requires efficient tracking of vaccination to assess coverage, safety, and efficacy of these vaccines. Paper-based immunization records are still being used in most low and middle-income countries. Adequate Electronic Logistic Management Information Systems, Immunization Registries and Records are crucial for proper data collection and analysis, and for making better decisions at an individual and at a population level. In this paper we share our experience in the redesign of an interoperable immunization record to track vaccination, including the recently developed vaccines for the novel coronavirus SARS-CoV-2 (COVID-19).

Clinical documentation in healthcare institutions is one of the daily tasks that consumes most of the time for those involved. The adoption of mobile devices in medical practice increases efficiency among healthcare professionals. We describe the design and evaluation of an automatic speech recognition system that enables the transcription of audio to text of clinical notes in a mobile environment. Our system achieved 94.1% word accuracy when evaluated on pediatrics, internal medicine and surgery services.

Accurate patient problem lists in Electronic Health Records (EHRs) are valuable tools for improving the quality of care, communication among professionals, facilitating research, quality measurement and the implementation of clinical decision support systems. However, problem lists are frequently inaccurate and out-of-date, and use varies widely across providers. These deficiencies limit problem list benefits. We decided to check if accuracy of problem lists-assessed at a granular level of detail-registered in EHRs depended on the specialty of the physician (primary care providers vs. specialists), and in the event that such differences did occur, whether or not accuracy had also been affected by the work environment. By using registered problems and taking the generated clinical document, we designed a cross-sectional survey following the guidelines of the Clinical Document Architecture standard. Problems registered by primary care providers have a higher level of accuracy than those registered by specialists in all settings considered (emergency unit, inpatient and outpatient). The work environment also significantly affects the accuracy level of problems registered.

This paper describes the steps followed in the creation of a local Interface Terminology to SNOMED CT (as reference terminology) with a strong focus on user acceptability. The resulting list of terms is used for clinical data input by physicians and nurses at the Hospital Italiano in Buenos Aires, Argentina. Description includes data model, mappings to SNOMED CT and classifications, subsets definitions and extensibility mechanisms. The Interface Terminology is currently used in the recording of diagnosis and procedures in inpatient discharge summaries and its coverage is improving from user feedback. Its current size is 24,800 concepts, 67% of them needed post-coordination for appropriate semantic representation, due to a very flexible policy that allows the use of any number of modifiers on concepts.

As the medical informatics field evolves, new functions appear as the focus of interest; a more advanced management of terminology is one of them. Using comprehensive and detailed terminology to represent clinical rules in computer systems, associated with patient information, would allow clinical software to provide patient specific recommendations or alerts. In order to uniform data collection through our HIS, and lay the foundations for future clinical decision support systems, we decided to move from our previous classification-based medical record into new terminology services built around Snomed CT, Spanish Language Version. This paper describes the characteristics of our Terminology Server. The most important achievements of our new terminology system are the centralization of knowledge representation, using a much more detailed terminology system. Clinical data entered at any place of the institution and level of care, is represented uniformly through the whole health information system.