Leveraging IT, AI and Data Science for Healthcare Beyond the Hospital: Learning from Scientific, Operational, and Business Perspectives
Permanent URI for this collection
Browse
Recent Submissions
1 - 6 of 6
-
ItemContext Changes and the Performance of a Learning Human-in-the-loop System: A Case Study of Automatic Speech Recognition Use in Medical Transcription( 2023-01-03)The paper presents how organizational practices enable the improvement and maintenance of task performance in a learning human-in-the-loop system exposed to a wide range of context changes. We investigate how the case company tripled the efficiency of medical transcribers by leveraging its machine learning-based automatic speech recognition technology. We find that the focal system operated across stable, drifting, and jumping contexts. Despite changes, it continued to improve or maintained performance thanks to two sets of organizational practices aligning it with the context: extending and refining. This paper makes two key contributions: It shows the importance of considering context changes in the design and operation of learning human-in-the-loop systems. Our empirical findings help with resolving some contradictory outcomes of the recent conceptual work. Secondly, we show that context alignment practices are situated at the sociotechnical system level and, thus, are not just technical solution nor can be detached from social elements.
-
ItemFunctional Assessment Data: Current Status of Federal Initiatives to Support Interoperability among Post Acute Care Settings( 2023-01-03)Health information needed along the transitions in care includes functional status such as self-care abilities assessments. Despite current federal efforts to support interoperability of functional status data, gaps still exist. Functional status assessments are included in data collection instruments widely used in four post acute care (PAC) settings, with each type of setting using a different standard instrument. These various instruments lack a shared standard for the content (meaning) of functional assessment items, necessitating mapping to a standard data terminology. Analysis indicates complete LOINC representation and incomplete SNOMED representation among functional status items and instruments. The new U.S. Core Data for Interoperability (USCDI) data standard has not included functional status in the next version to be adopted due in part to insufficiently defined use cases. The Post-Acute Care Interoperability Workgroup (PACIO) produced a FHIR implementation guide for functional status based on a use case. Gaps persist in PAC interoperability adoption.
-
Item
-
ItemExploration of Sleep Events in the Latent Space of Variational Autoencoders on a Breath-by-Breath Basis( 2023-01-03)In this exploratory paper, we attempt to address a growing demand for unsupervised machine learning techniques on sleep data by applying a variational autoencoder on respiratory sleep data on a breath-by-breath basis. We transform respiratory signals into a latent representation and cluster them together using KMeans clustering. We calculate the cluster preference of scored events and attempt to explain their position in the latent space. We show that a variational autoencoder can accurately reconstruct three respiratory signals from individual breaths despite being sampled through a latent dimension 384 times smaller than the input data. Our results also indicate that respiratory events in particular show a tendency to cluster together in the latent space despite a purely unsupervised learning approach. Finally, we lay the groundwork for future work made possible in this paper.
-
ItemTowards a Deeper Understanding of Sleep Stages through their Representation in the Latent Space of Variational Autoencoders( 2023-01-03)Artificial neural networks show great success in sleep stage classification, with an accuracy comparable to human scoring. While their ability to learn from labelled electroencephalography (EEG) signals is widely researched, the underlying learning processes remain unexplored. Variational autoencoders can capture the underlying meaning of data by encoding it into a low-dimensional space. Regularizing this space furthermore enables the generation of realistic representations of data from latent space samples. We aimed to show that this model is able to generate realistic sleep EEG. In addition, the generated sequences from different areas of the latent space are shown to have inherent meaning. The current results show the potential of variational autoencoders in understanding sleep EEG data from the perspective of unsupervised machine learning.
-
ItemPredicting Remote Monitoring Patients’ Non-compliance Behavior Through App-mediated Communications( 2023-01-03)Remote patient monitoring (RPM) has been widely used for monitoring patients’ health and tracking their behavior outside the traditional healthcare setting. One important behavior to understand is patients’ compliance with medical advice and treatment regimes. Existing methods detect non-compliance based on health parameters i.e., weight and vital signs, which can only be identified by the deterioration in health conditions. This study proposes an RPM system artifact to record patients’ feelings and concerns through short messages; these messages are used to develop a non-compliance prediction model. A prototype of the design artifact was implemented and tested with chronic patients taking home hemodialysis. Our model revealed that the counts of messages recorded are related to non-compliance behavior, and the negative emotions depicted in the messages implied a higher likelihood of non-compliance. Our study demonstrated the feasibility of understanding patients’ status based on non-health parameters and provided a way to enhance RPM for patients outside the hospital settings.