Delivering Health Intelligence For Healthcare Services

Abstract

The systems barrier for clinical information interoperability and standards has now evolved from a technology barrier to a semantic barrier. The processes to gather clinical data and to build clinical information and knowledge cannot be fully implemented, owing to semantic dissonances and limited data normalization. According to [1], "Just over a half of entered codes were appropriate for a given scenario and about a quarter were omitted." This is a significant data and financial gap for healthcare provision. Huge amount of addition to the financial cost, lack of data integration and loss of information affects the ability to maintain standards in clinical care delivery and patient outcomes.This paper proposes that the solution to these issues is an augmented network of clinical note taking, where coding is automatically generated by an AI system as clinicians write their clinical notes. The system (AI-KEN) offers enhanced web support that is integrated to local clinical systems, whereby clinical notes are prompted by suggested predictive text options in real time. The anticipated benefits include reducing financial loss for acute services, support for clinical standard maintenance and enhanced advancements for clinical practice and research in real time.

Recent Advances of Artificial Intelligence in Healthcare: A Systematic Literature Review

“…In order to attain a high level of accuracy, the authors suggest making use of search methods for feature selection. Murray et al [84] investigate potential solutions based on machine learning to address the semantic issues that are now being faced by healthcare practitioners.…”

“…According to the authors, they conclude with a strategy that combines health coaching with AI in order to assist patients in more efficiently managing their chronic diseases and reduce the number of visits that are not essential. This system, which is very similar to the one proposed by Murray et al [84], is made up of sensors that are able to collect biometric data, AI models that are able to create insights about health issues, and visual analytics tools that are able to display relevant data in graphical and textual formats.…”

“…AI applications and IT tools are used by organizations in order to cut costs [19,20,24,88], detect fraud [17,89], improve performance [10,27,51], and provide workflow assistance [14,37,53,90]. For example, Murray et al [84] highlight the challenge of automatically extracting knowledge using medical IS. This difficulty is mostly due to the limited level of data standardization and integration that is present in these systems.…”

“…Due to the ethical implications of such data, it can be challenging to gather data at times, particularly data pertaining to patients. If some classification and clustering methods are applied to a very small amount of data, the results may be quite accurate; nevertheless, this may not be practical or useful [60,64,84]. For AI methods to work well, the data must first be preprocessed.…”

“…In particular, natural language processing must be performed extensively on text data before they can be used. One of the most difficult issues in medical data processing is the need to integrate several forms of data using the same algorithm at times [60,64,84]. This is one of the reasons why there are so many different types of algorithms.…”