Human Interpretation Research Articles

On the Suitability of Process Mining for Enhancing Transparency of Blockchain Applications

AbstractBlockchain technology is known for its transparency properties due to its publicly available, immutable data. Yet, as data availability does not inherently ensure transparency, further analytical methods may be required for human interpretation of data traces. Process mining has emerged as a popular toolbox for understanding processes and how they are executed in practice. The paper studies process mining as a method to enhance the transparency of blockchain data. To this end, two popular Ethereum applications were analyzed using process mining: the prediction and betting marketplace Augur and the network marketing platform Forsage. Observations from the process-mining analyses are used to discuss if process mining can serve as a method to establish transparency of a blockchain. For both applications, new insights are generated for usage scenarios such as application redesign, security analysis, user behavior analysis, and revealing blind spots in Augur’s and Forsage’s documentation. The paper concludes that there is evidence that process mining can serve as a method to enhance transparency in blockchains at the cost of technical setup and knowledge acquisition.

Deep Learning-Driven Sign Language Recognition: A Multimodal Approach for Gesture-to-Text Translation Using CNN-RNN Architectures

Sign language serves as a vital communication tool for individuals with hearing and speech impairments. However, a significant barrier exists when others do not possess a strong understanding of sign language often requiring interpreters to facilitate communication. To reduce the reliance on human interpreters, this research aims to develop an intelligent system capable of recognizing and translating sign language gestures into meaningful grammatically correct sentences. The proposed system processes both images and videos to interpret gestures, converting them into text that can be translated into any desired language. The system leverages state-of-the-art deep learning techniques, such as Convolutional Neural Networks (CNNs) for identifying key features in images and Recurrent Neural Networks (RNNs) for understanding temporal sequences in video data. By employing these advanced neural networks, the model is able to comprehend hand movements, facial expressions and other non-verbal cues to construct coherent sentences. Additionally, the system integrates natural language processing (NLP) to refine the output, ensuring the resulting sentences are grammatically correct. Our approach addresses common challenges such as differentiating between subtle hand gestures and reducing the impact of environmental noise in images. This solution holds the potential to significantly enhance communication for the hearing and speech impairedffering an efficient interpreter-free method of translating sign language into widely spoken languages

Reconstructing a georeferenced inventory of flooding hazards in Raoping, Guangdong province, China, from 1492 to 1985

ABSTRACT In ancient times, there is a scarcity of instrumental water flow data, which challenges the hydrological science community to understand the evolution of flood risk under the changing climate over the centennial scale. Based on the historical records of flood events, the first spatio-temporal database of flood risk occurrence was established in Raoping county, Guangdong, China from 1492 to 1985 (a 494-year period), with intensive human interpretation and local investigation. Specifically, the spatially explicit flood events of the river network were provided for each flood episode. A detailed analysis of the database shows a high frequency of flood risk in summer and early autumn. Specifically, a significant (p < 0.05) increase has been found in flood risks since 1960. However, flooding decreased significantly in recent decades due to meteorological and hydrological factors, as well as the population density and migration during the 500-year period. A spatial clustering of flood events in the northern and southern parts is also confirmed, which shows an impact of population dynamics on a centennial scale. Such methods can be a reference for establishing China's flooding database for ancient periods, promoting a better understanding of natural hazards and associated human behaviors under the context of long-term climate change.

Toward comprehensive short utterances manipulations detection in videos

AbstractIn a landscape increasingly populated by convincing yet deceptive multimedia content generated through generative adversarial networks, there exists a significant challenge for both human interpretation and machine learning algorithms. This study introduces a shallow learning technique specifically tailored for analyzing visual and auditory components in videos, targeting the lower face region. Our method is optimized for ultra-short video segments (200-600 ms) and employs wavelet scattering transforms for audio and discrete cosine transforms for video. Unlike many approaches, our method excels at these short durations and scales efficiently to longer segments. Experimental results demonstrate high accuracy, achieving 96.83% for 600 ms audio segments and 99.87% for whole video sequences on the FakeAVCeleb and DeepfakeTIMIT datasets. This approach is computationally efficient, making it suitable for real-world applications with constrained resources. The paper also explores the unique challenges of detecting deepfakes in ultra-short sequences and proposes a targeted evaluation strategy for these conditions.

Contextualizing Human Rights in Multicultural Environments

This paper explores the intricate relationship between socio-cultural dynamics and human rights in multicultural societies. It examines how norms, belief systems, and power structures shape the perception and exercise of these universal principles. The study employs a qualitative research design, specifically using the Cultural Relativism Framework, to investigate how cultural norms and values influence the interpretation and application of human rights. Through the analysis of literature and case studies, the paper elucidates the interaction between cultural context and human rights interpretations. It acknowledges that human rights are not uniform but are influenced by diverse cultural contexts. Key findings underscore the significance of multiculturalism in human rights interpretation, the impact of social, cultural, and political contexts, the intersection of values, beliefs, and customs, and the roles of states, civil society, and international organizations in shaping human rights discourse. The paper advocates for educational initiatives, intercultural dialogue, inclusive policymaking, and legal reforms to promote and protect human rights in diverse societies. It recommends enhancing human rights education, facilitating intercultural exchange, empowering civil society, addressing discrimination and inequality, and integrating customary laws with formal legal systems. By fostering mutual respect and understanding, societies can uphold human rights and cultural diversity, enabling peaceful, equitable coexistence. This assertion is based on empirical observations and scientific principles that highlight the importance of mutual understanding in preserving human rights and promoting cultural diversity. This paper contributes to a deeper understanding of human rights, advocating for a balance between universal principles and cultural specificities.

A “Region-Specific Model Adaptation (RSMA)”-Based Training Data Method for Large-Scale Land Cover Mapping

An accurate and historical land cover monitoring dataset for Alaska could provide fundamental information for a range of studies, such as conservation habitats, biogeochemical cycles, and climate systems, in this distinctive region. This research addresses challenges associated with the extraction of training data for timely and accurate land cover classifications in Alaska over longer time periods (e.g., greater than 10 years). Specifically, we designed the “Region-Specific Model Adaptation (RSMA)” method for training data. The method integrates land cover information from the National Land Cover Database (NLCD), LANDFIRE’s Existing Vegetation Type (EVT), and the National Wetlands Inventory (NWI) and machine learning techniques to generate robust training samples based on the Anderson Level II classification legend. The assumption of the method is that spectral signatures vary across regions because of diverse land surface compositions; however, despite these variations, there are consistent, collective land cover characteristics that span the entire region. Building upon this assumption, this research utilized the classification power of deep learning algorithms and the generalization ability of RSMA to construct a model for the RSMA method. Additionally, we interpreted existing vegetation plot information for land cover labels as validation data to reduce inconsistency in the human interpretation. Our validation results indicate that the RSMA method improved the quality of the training data derived solely from the NLCD by approximately 30% for the overall accuracy. The validation assessment also demonstrates that the RSMA method can generate reliable training data on large scales in regions that lack sufficient reliable data.

A-140 Reducing cell count bias with the GloCyte® Automated Cell Counter for CSF: an observational study at Loyola University Medical Center

Abstract Background Accurate STAT cerebrospinal fluid (CSF) cell counts are critical in supporting diagnoses of meningitis, malignancy, demyelinating disease, and hemorrhage. Limitations of automated analyzers to achieve acceptable linearity for the clinically relevant range has maintained manual CSF cell counting as the preferred method. Manual cell counting, performed with a hemocytometer and bright field microscope, is time-consuming, requires specialized skills and depends on human interpretation. This subjectivity inherently introduces the potential for non-random error in the enumeration of both total nucleated cells (TNC) and red blood cells in CSF by manual methods. The purpose of this study was to evaluate and compare the level of bias inherent in manual determination (TNC) in CSF as compared to the GloCyte Automated Cell Counter. Methods Result bias data for CSF specimens processed in the hematology lab at Loyola University Medical Center, a quaternary care 547-bed academic medical center in Maywood, IL, was analyzed for manual cell counting using a Neubauer-improved hemocytometer and GloCyte, an FDA-cleared cell counting device employing fluorescence technology to automate RBC and TNC counts in under 5 minutes, with linearity down to 0 cells/µL. Result bias was determined by defining the “expected result” for both GloCyte and hemocytometer as the initial cell count determined at a 1:5 dilution of a freshly-collected CSF sample and calculating the expected cell count for subsequent serial dilutions of 1:10, 1:20, 1:50, 1:100 for each method. The actual TNC count and the disparity of the actual result from the expected result was recorded as bias. Results Data are summarized below in the Table, and indicate a lower bias for GloCyte compared to manual counts across the range of dilutions. Conclusions This study demonstrates that GloCyte can be implemented as a means of effectively mitigating the reporting bias of CSF TNC counts performed with a Neubauer-improved hemocytometer.

Deep Learning Method Applied to Autonomous Image Diagnosis for Prick Test.

The skin prick test (SPT) is used to diagnose sensitization to antigens. This study proposes a deep learning approach to infer wheal dimensions, aiming to reduce dependence on human interpretation. A dataset of SPT images (n = 5844) was used to infer a convolutional neural network for wheal segmentation (ML model). Three methods for inferring wheal dimensions were evaluated: the ML model; the standard protocol (MA1); and approximation of the area as an ellipse using diameters measured by an allergist (MA2). The results were compared with assisted image segmentation (AIS), the most accurate method. Bland-Altman analysis, distribution analyses, and correlation tests were applied to compare the methods. This study also compared the percentage deviation among these methods in determining the area of wheals with regular geometric shapes (n = 150) and with irregular shapes (n = 150). The Bland-Altman analysis showed that the difference between methods was not correlated with the absolute area. The ML model achieved a segmentation accuracy of 85.88% and a strong correlation with the AIS method (ρ = 0.88), outperforming all other methods. Additionally, MA1 showed significant error (13.44 ± 13.95%) for pseudopods. The ML protocol can potentially automate the reading of SPT, offering greater accuracy than the standard protocol.

Interpretable few-shot learning with online attribute selection

Few-shot learning (FSL) presents a challenging learning problem in which only a few samples are available for each class. Decision interpretation is more important in few-shot classification due to a greater chance of error compared to traditional classification. However, the majority of the previous FSL methods are black-box models. In this paper, we propose an inherently interpretable model for FSL based on human-friendly attributes. Previously, human-friendly attributes have been utilized to train models with the potential for human interaction and interpretability. However, such approaches are not directly extendible to the few-shot classification scenario. Moreover, we propose an online attribute selection mechanism to effectively filter out irrelevant attributes in each episode. The attribute selection mechanism improves accuracy and helps with interpretability by reducing the number of attributes that participate in each episode. We further propose a mechanism that automatically detects the episodes where the pool of available human-friendly attributes is insufficient, and subsequently augments it by engaging some learned unknown attributes. We demonstrate that the proposed method achieves results on par with black-box few-shot learning models on four widely used datasets. We also empirically evaluate the level of decision alignment between different models and human understanding and show that our model outperforms the comparison methods based on this criterion.

Evaluating a Natural Language Processing-Driven, AI-Assisted International Classification of Diseases, 10th Revision, Clinical Modification, Coding System for Diagnosis Related Groups in a Real Hospital Environment: Algorithm Development and Validation Study.

International Classification of Diseases codes are widely used to describe diagnosis information, but manual coding relies heavily on human interpretation, which can be expensive, time consuming, and prone to errors. With the transition from the International Classification of Diseases, Ninth Revision, to the International Classification of Diseases, Tenth Revision (ICD-10), the coding process has become more complex, highlighting the need for automated approaches to enhance coding efficiency and accuracy. Inaccurate coding can result in substantial financial losses for hospitals, and a precise assessment of outcomes generated by a natural language processing (NLP)-driven autocoding system thus assumes a critical role in safeguarding the accuracy of the Taiwan diagnosis related groups (Tw-DRGs). This study aims to evaluate the feasibility of applying an International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), autocoding system that can automatically determine diagnoses and codes based on free-text discharge summaries to facilitate the assessment of Tw-DRGs, specifically principal diagnosis and major diagnostic categories (MDCs). By using the patient discharge summaries from Kaohsiung Medical University Chung-Ho Memorial Hospital (KMUCHH) from April 2019 to December 2020 as a reference data set we developed artificial intelligence (AI)-assisted ICD-10-CM coding systems based on deep learning models. We constructed a web-based user interface for the AI-assisted coding system and deployed the system to the workflow of the certified coding specialists (CCSs) of KMUCHH. The data used for the assessment of Tw-DRGs were manually curated by a CCS with the principal diagnosis and MDC was determined from discharge summaries collected at KMUCHH from February 2023 to April 2023. Both the reference data set and real hospital data were used to assess performance in determining ICD-10-CM coding, principal diagnosis, and MDC for Tw-DRGs. Among all methods, the GPT-2 (OpenAI)-based model achieved the highest F1-score, 0.667 (F1-score 0.851 for the top 50 codes), on the KMUCHH test set and a slightly lower F1-score, 0.621, in real hospital data. Cohen κ evaluation for the agreement of MDC between the models and the CCS revealed that the overall average κ value for GPT-2 (κ=0.714) was approximately 12.2 percentage points higher than that of the hierarchy attention network (κ=0.592). GPT-2 demonstrated superior agreement with the CCS across 6 categories of MDC, with an average κ value of approximately 0.869 (SD 0.033), underscoring the effectiveness of the developed AI-assisted coding system in supporting the work of CCSs. An NLP-driven AI-assisted coding system can assist CCSs in ICD-10-CM coding by offering coding references via a user interface, demonstrating the potential to reduce the manual workload and expedite Tw-DRG assessment. Consistency in performance affirmed the effectiveness of the system in supporting CCSs in ICD-10-CM coding and the judgment of Tw-DRGs.

Text mining of verbal autopsy narratives to extract mortality causes and most prevalent diseases using natural language processing

Verbal autopsy (VA) narratives play a crucial role in understanding and documenting the causes of mortality, especially in regions lacking robust medical infrastructure. In this study, we propose a comprehensive approach to extract mortality causes and identify prevalent diseases from VA narratives utilizing advanced text mining techniques, so as to better understand the underlying health issues leading to mortality. Our methodology integrates n-gram-based language processing, Latent Dirichlet Allocation (LDA), and BERTopic, offering a multi-faceted analysis to enhance the accuracy and depth of information extraction. This is a retrospective study that uses secondary data analysis. We used data from the Agincourt Health and Demographic Surveillance Site (HDSS), which had 16338 observations collected between 1993 and 2015. Our text mining steps entailed data acquisition, pre-processing, feature extraction, topic segmentation, and discovered knowledge. The results suggest that the HDSS population may have died from mortality causes such as vomiting, chest/stomach pain, fever, coughing, loss of weight, low energy, headache. Additionally, we discovered that the most prevalent diseases entailed human immunodeficiency virus (HIV), tuberculosis (TB), diarrhoea, cancer, neurological disorders, malaria, diabetes, high blood pressure, chronic ailments (kidney, heart, lung, liver), maternal and accident related deaths. This study is relevant in that it avails valuable insights regarding mortality causes and most prevalent diseases using novel text mining approaches. These results can be integrated in the diagnosis pipeline for ease of human annotation and interpretation. As such, this will help with effective informed intervention programmes that can improve primary health care systems and chronic based delivery, thus increasing life expectancy.

Strategic Framework for Human-Centric AI Governance: Navigating Ethical, Educational, and Societal Challenges

Abstract: Humanity is at the precipice of a profound transformation, propelled by the rapid evolution of artificial intelligence (AI). Once confined to the realm of science fiction, AI now permeates every aspect of daily life, promising significant changes to our social, economic, and personal spheres. As AI systems continue to advance and surpass human capabilities in various domains, they present unparalleled opportunities as well as formidable challenges. This composition explores the dualistic nature of AI, acknowledging its capacity to either augment or disrupt our existence. By examining AI's current state, potential risks, and ethical considerations, we emphasize the pressing need for a comprehensive framework to govern its development and integration. We propose a multifaceted framework built on five fundamental principles: human centered design, transparency and interpretability, accountability and governance, education and upskilling, and inclusive and equitable access. These principles are designed to ensure that AI systems enhance human capabilities, promote transparency, uphold ethical standards, and provide equitable access to the benefits of technology. Through in-depth case studies, we illustrate the practical application of this framework, showcasing AI's impact across various industries while emphasizing the importance of addressing ethical and societal challenges. This composition calls for a collective effort among governments, corporations, and individuals to navigate the rapid expansion of AI responsibly, fostering a future where AI and humanity coexist harmoniously. The ultimate goal is to harness the potential of AI to revitalize society, ensuring that technological progress elevates human dignity and collective well-being instead of rendering us obsolete.

Quality Assessment of Rice Using Convolution Neural Networks and Other Machine Learning Techniques- A Comparative Study

Rice holds its position as the most widely cultivated crop worldwide, its demand steadily rising alongside population growth. The precise identification of grains, especially wheat and rice, holds significant importance as their ultimate utilization depends on the quality of grains prior to processing. Traditionally, grain identification tasks have been predominantly manual, relying on experienced grain inspectors and consuming considerable time. However, this manual classification process is susceptible to variations influenced by individual perception, given the subjective nature of human image interpretation. Consequently, there is an urgent need for an automated recognition system capable of accurately identifying grains under diverse environmental conditions, necessitating the application of digital image processing techniques. In this study, we focus on grading five distinct varieties of rice based on their quality, employing a range of convolution neural networks (CNN) namely, Efficientnetb0, Googlenet, MobileNetV2, Resnet50, Resnet101, and ShuffleNet. The performance of CNN towards identification and grading of rice grain is also compared to that of other parametric and Non-parametric classifiers namely, Linear Discriminant Analysis (LDA), K-Nearest Neighbor (K-NN), Naive Bayes (NB), and Back Propagation Neural Network (BPNN) using Gray Level Co-occurrence Matrix (GLCM) and Gray Level Run Length Matrix (GLRLM) based texture features. The image dataset comprises five grades of rice, each containing 100 images, resulting in a comprehensive collection of 500 samples for analysis. It is observed that, Convolution neural networks can grade five different qualities of rice with highest accuracy of 64.4% in case of GoogleNet. Results show that, rice grading using texture features performs better with highest accuracy of 99.2% (using GLCM) and 93.4% (using GLRLM).

Determining research priorities using machine learning

We summarize our exploratory investigation into whether Machine Learning (ML) techniques applied to publicly available professional text can substantially augment strategic planning for astronomy. We find that an approach based on Latent Dirichlet Allocation (LDA) using content drawn from astronomy journal papers can be used to infer high-priority research areas. While the LDA models are challenging to interpret, we find that they may be strongly associated with meaningful keywords and scientific papers which allow for human interpretation of the topic models.Significant correlation is found between the results of applying these models to the previous decade of astronomical research (“1998–2010” corpus) and the contents of the Science Frontier Panels report which contains high-priority research areas identified by the 2010 National Academies’ Astronomy and Astrophysics Decadal Survey (“DS2010” corpus). Significant correlations also exist between model results of the 1998–2010 corpus and the submitted whitepapers to the Decadal Survey (“whitepapers” corpus). Importantly, we derive predictive metrics based on these results which can provide leading indicators of which content modeled by the topic models will become highly cited in the future. Using these identified metrics and the associations between papers and topic models it is possible to identify important papers for planners to consider.A preliminary version of our work was presented by Thronson et al. (2021) and Thomas et al. (2022).

Thermodynamics-inspired explanations of artificial intelligence

In recent years, predictive machine learning models have gained prominence across various scientific domains. However, their black-box nature necessitates establishing trust in them before accepting their predictions as accurate. One promising strategy involves employing explanation techniques that elucidate the rationale behind a model’s predictions in a way that humans can understand. However, assessing the degree of human interpretability of these explanations is a nontrivial challenge. In this work, we introduce interpretation entropy as a universal solution for evaluating the human interpretability of any linear model. Using this concept and drawing inspiration from classical thermodynamics, we present Thermodynamics-inspired Explainable Representations of AI and other black-box Paradigms, a method for generating optimally human-interpretable explanations in a model-agnostic manner. We demonstrate the wide-ranging applicability of this method by explaining predictions from various black-box model architectures across diverse domains, including molecular simulations, text, and image classification.

A multicentre study to evaluate the diagnostic performance of a novel CAD software, DecXpert, for radiological diagnosis of tuberculosis in the northern Indian population

Tuberculosis (TB) is the leading cause of mortality among infectious diseases globally. Effectively managing TB requires early identification of individuals with TB disease. Resource-constrained settings often lack skilled professionals for interpreting chest X-rays (CXRs) used in TB diagnosis. To address this challenge, we developed “DecXpert” a novel Computer-Aided Detection (CAD) software solution based on deep neural networks for early TB diagnosis from CXRs, aiming to detect subtle abnormalities that may be overlooked by human interpretation alone. This study was conducted on the largest cohort size to date, where the performance of a CAD software (DecXpert version 1.4) was validated against the gold standard molecular diagnostic technique, GeneXpert MTB/RIF, analyzing data from 4363 individuals across 12 primary health care centers and one tertiary hospital in North India. DecXpert demonstrated 88% sensitivity (95% CI 0.85–0.93) and 85% specificity (95% CI 0.82–0.91) for active TB detection. Incorporating demographics, DecXpert achieved an area under the curve of 0.91 (95% CI 0.88–0.94), indicating robust diagnostic performance. Our findings establish DecXpert's potential as an accurate, efficient AI solution for early identification of active TB cases. Deployed as a screening tool in resource-limited settings, DecXpert could enable early identification of individuals with TB disease and facilitate effective TB management where skilled radiological interpretation is limited.

Complexity Analysis Using Graph Models for Conflict Resolution for Autonomous Ships in Complex Situations

Abstract Maritime autonomous surface ships (MASSs) will reshape the fast-evolving ecosystem for their attractive socio-economic benefits and potential to improve safety. However, their new systems and technology need thorough verifications to identify unintended components of risk. The interaction between MASS cyber-physical systems and the existing regulatory framework is currently unpredictable; artificial intelligence-powered intelligent situation awareness and autonomous navigation algorithms must safely and efficiently adhere to the regulations which are only designed for human interpretation without MASSs consideration. This paper contributes to algorithmic regulations and particularly algorithmic COLREGs in real-world MASS applications. It focuses on codifying COLREGs into a machine-executable system applicable to MASSs, then analyzing their performance in dynamic and mixed interactions between multiple vessels in complex scenarios. Based on fullest pairwise COLREGs criteria, this paper considers decision-making (DM) and complexity analysis in multi-collision-conflict scenarios. Complexity influential factors are an interplay between the characteristics of COLREGs, traffic scenarios, MASS interactions, and the environment. Participant vessels are the decision-makers forming a decentralized uncertain DM process, casted into a multi-participant multi-conflict multicriteria DM problem. This is tackled through the technique of graph models for conflict resolution, using risk graph models and fuzzy preferences over alternative collision-avoidance states. This paper conducts a comprehensive analysis of DM and navigational complexity; we develop novel complexity estimation scores, tools for complexity monitoring for human intervention, and spatial analysis of traffic complexity for geo-intelligent MASSs deployment and operation planning. The presented work is validated on a database of historical scenarios extracted from multiple data sources.

A Missing Link on Semantic Interoperability.

Syntactic interoperability among health ICT systems is well-established, but achieving semantic interoperability requires more than just exchanging codes. We present a pragmatic, safe, and effective approach towards an ambitious goal: enabling any software to process a critical mass of routine clinical information in a replicable manner across various systems and local contexts. We advocate for the use of reliable, parsimonious coding to handle the most replicable aspects of data processing for routine patient information, while reserving the intricate interpretation of individual patient data nuances for skilled professionals, possibly supported by Artificial Intelligence tools. We suggest coping with routine tasks by focusing on a limited set of a few thousand data elements, named the 'Clinical Documentation Kernel' (CDK). This approach will provide direct benefits to users and assist in the human interpretation of other patient information. Our preliminary study focuses on the 'primitives' and 'qualifiers' that bring the highest value to the health ecosystem in various authoritative scenarios in the field of diabetes.

Feasibility of Artificial Intelligence Powered Adverse Event Analysis: Using a Large Language Model to Analyze Microwave Ablation Malfunction Data.

Objectives: Determine if a large language model (LLM, GPT-4) can label and consolidate and analyze interventional radiology (IR) microwave ablation device safety event data into meaningful summaries similar to humans. Methods: Microwave ablation safety data from January 1, 2011 to October 31, 2023 were collected and type of failure was categorized by human readers. Using GPT-4 and iterative prompt development, the data were classified. Iterative summarization of the reports was performed using GPT-4 to generate a final summary of the large text corpus. Results: Training (n = 25), validation (n = 639), and test (n = 79) data were split to reflect real-world deployment of an LLM for this task. GPT-4 demonstrated high accuracy in the multiclass classification problem of microwave ablation device data (accuracy [95% CI]: training data 96.0% [79.7, 99.9], validation 86.4% [83.5, 89.0], test 87.3% [78.0, 93.8]). The text content was distilled through GPT-4 and iterative summarization prompts. A final summary was created which reflected the clinically relevant insights from the microwave ablation data relative to human interpretation but had inaccurate event class counts. Conclusion: The LLM emulated the human analysis, suggesting feasibility of using LLMs to process large volumes of IR safety data as a tool for clinicians. It accurately labelled microwave ablation device event data by type of malfunction through few-shot learning. Content distillation was used to analyze a large text corpus (>650 reports) and generate an insightful summary which was like the human interpretation.

Implementation of a digital distress detection system in palliative care: qualitative data on perspectives of a multiprofessional palliative care team

BackgroundDigital health technologies such as sensor systems are intended to support healthcare staff in providing adequate patient care. In the Department of Palliative Medicine (University Medical Center Freiburg), we developed and implemented a noninvasive, bed-based sensor system in a pilot study. The aim was to detect distress in patients who were no longer able to express themselves by monitoring heart and respiratory rates, vocalizations, and movement measurements. The sensor system was intended to supplement standard care, which generally cannot guarantee constant monitoring. As there is a lack of data on how healthcare professionals experience such a techno-digital innovation, the aim of this study was to explore how the multiprofessional palliative care team who piloted the sensor system perceived its potential benefits and limitations, and how they experienced the broader context of healthcare technology and research in palliative care.MethodsWe conducted a qualitative interview study with 20 members of the palliative care team and analyzed the recorded, verbatim transcribed interviews using qualitative content analysis.ResultsThe sensor system was described as easy to use and as helpful support for patients, care staff, and relatives, especially against the backdrop of demographic change. However, it could not replace human interpretation of stress and subsequent treatment decisions: this remained the expertise of the nursing staff. A potential reduction in personnel was expected to be a risk of a digital monitoring system. The special conditions of research and digital health technologies in an end-of-life context also became clear. Specifically, healthcare staff were open to health technologies if they benefited the patient and were compatible with professional nursing and/or palliative care attitudes. Additionally, a patient-protective attitude and possible interprofessional differences in priorities and the resulting challenges for the team became apparent.ConclusionsA potential digital solution for distress monitoring was considered useful by palliative care practitioners. However, interprofessional differences and compatibility with existing palliative care practices need to be considered before implementing such a system. To increase user acceptability, the perspectives of healthcare professionals should be included in the implementation of technological innovations in palliative care.