Normalization of medical concepts to an ontology is a key aspect of the natural language processing of biomedical text. It enables the mapping of medical expressions to standardized ontology terms and their identifiers, thereby enhancing the interoperability and computability of medical concepts. Although large language models (LLMs) can identify and standardize medical terms, they may struggle to accurately map ontology terms to their corresponding ontology identifiers. These challenges arise from the stochastic nature of LLMs, their limited exposure to uncommon ontology identifiers during training, and their lack of an integrated lookup mechanism. We generated test sets of synthetic terms to assess normalization performance by both zero-shot prompted and retrieval-augmented generation (RAG) prompted methods across two ontologies (Human Phenotype Ontology and Gene Ontology) and three LLMs (GPT-4o, LLaMA 3.3 70B, and Phi-4). To ensure a calibrated and fair evaluation of normalization, the test set was balanced along two axes: (1) term prevalence in biomedical literature, as estimated by PubMed Central frequency counts, and (2) semantic proximity to ontology terms, as assessed by cosine similarity of BioBERT embeddings. Our results demonstrate that RAG consistently outperforms zero-shot prompting, particularly on low-prevalence terms that are infrequently encountered in the biomedical literature. This highlights the value of RAG in compensating for gaps in model exposure to uncommon medical concepts. We demonstrate that a synthetic test set can be a valuable tool for evaluating biomedical term normalization across LLMs.

Dimension reduction methods are used to visualize the output of unsupervised learning models when applied to complex data. These techniques improve interpretability by transforming a high-dimension space to a lower-dimension space (usually 2D or 3D). The results are typically viewed as 2D scatter plots, and class centroids may be added to increase interpretability. Although useful, the relationship of these class centroids to the underlying feature space remains opaque. The innovative aspect of this work is to create a strong link between the dimension-reduced space and the underlying high-dimension feature space by adding selected feature centroids to the 2D scatter plots. This approach simultaneously visualizes the centers for the classes and the features on the same 2D scatter plot. Since classes are often imbalanced, we provide a method to balance class sizes. We present an automated framework that performs a grid search to find the optimal dimension reduction parameters, balances the class sizes, uses an ensemble approach to find the most important features, and adds class centroids and selected feature centroids to 2D dimension-reduced plots. This is especially useful when applied to complex, feature-rich biomedical data, as addition of feature centroids to 2D scatter plots serve as landmarks for the previously featureless dimension-reduced space. The utility of this approach is demonstrated by its application to seven classes of neurogenetic diseases with 31 defining phenotypic features.

After traumatic brain injury (TBI), clinicians use the Glasgow Coma Scale (GCS) to classify patients by severity and radiologists use the Rotterdam score and the Marshall score to classify CT scans by severity. This work investigates a viable efficient low cost transfer learning model to use MRI images to predict GCS, Rotterdam, and Marshall severity class after TBI. The enhanced transfer learning model architecture integrates multiple fine-tuning steps in which a few layers of the pre-trained model is unfrozen in each iteration so that the neural network is able to learn layer by layer further aspects of the image for better classification. By conducting a thorough evaluation across multiple convolutional neural network (CNN) architectures, this work ascertains the sensitivity of CNN models in detecting anatomical changes presented in MRI images that are predictive of severity class after TBI. These models have potential to predict outcomes after TBI. We utilize both quantitative metrics and qualitative analysis to validate the clinical relevance of the models in predicting severity class on admission and outcome at 6 months. The residual network models (ResNet152 in particular) outperformed the other models in predicting initial severity class.

Deep Convolutional Neural Networks (CNNs) have become the go-to method for medical imaging classification on various imaging modalities for binary and multiclass problems. Deep CNNs extract spatial features from image data hierarchically, with deeper layers learning more relevant features for the classification application. Despite the high predictive accuracy, usability lags in practical applications due to the black-box model perception. Model explainability and interpretability are essential for successfully integrating artificial intelligence into healthcare practice. This work addresses the challenge of an explainable deep learning model for the prediction of the severity of Alzheimer’s disease (AD). AD diagnosis and prognosis heavily rely on neuroimaging information, particularly magnetic resonance imaging (MRI). We present a deep learning model framework that integrates a local data-driven interpretation method that explains the relationship between the predicted AD severity from the CNN and the input MR brain image. The deep explainer uses SHapley Additive exPlanation values to quantity the contribution of different brain regions utilized by the CNN to predict outcomes. We conduct a comparative analysis of three high-performing CNN models: DenseNet121, DenseNet169, and Inception-ResNet-v2. The framework shows high sensitivity and specificity in the test sample of subjects with varying levels of AD severity. We also correlated five key AD neurocognitive assessment outcome measures and the APOE genotype biomarker with model misclassifications to facilitate a better understanding of model performance.

Concussions, also known as mild traumatic brain injury (mTBI), are a growing health challenge. Approximately four million concussions are diagnosed annually in the United States. Concussion is a heterogeneous disorder in causation, symptoms, and outcome making precision medicine approaches to this disorder important. Persistent disabling symptoms sometimes delay recovery in a difficult to predict subset of mTBI patients. Despite abundant data, clinicians need better tools to assess and predict recovery. Data-driven decision support holds promise for accurate clinical prediction tools for mTBI due to its ability to identify hidden correlations in complex datasets. We apply a Locality-Sensitive Hashing model enhanced by varied statistical methods to cluster blood biomarker level trajectories acquired over multiple time points. Additional features derived from demographics, injury context, neurocognitive assessment, and postural stability assessment are extracted using an autoencoder to augment the model. The data, obtained from FITBIR, consisted of 301 concussed subjects (athletes and cadets). Clustering identified 11 different biomarker trajectories. Two of the trajectories (rising GFAP and rising NF-L) were associated with a greater risk of loss of consciousness or post-traumatic amnesia at onset. The ability to cluster blood biomarker trajectories enhances the possibilities for precision medicine approaches to mTBI.

For many neurological disorders, including traumatic brain injury (TBI), neuroimaging information plays a crucial role determining diagnosis and prognosis. TBI is a heterogeneous disorder that can result in lasting physical, emotional and cognitive impairments. Magnetic Resonance Imaging (MRI) is a non-invasive technique that uses radio waves to reveal fine details of brain anatomy and pathology. Although MRIs are interpreted by radiologists, advances are being made in the use of deep learning for MRI interpretation. This work evaluates a deep learning model based on a residual learning convolutional neural network that predicts TBI severity from MR images. The model achieved a high sensitivity and specificity on the test sample of subjects with varying levels of TBI severity. Six outcome measures were available on TBI subjects at 6 and 12 months. Group comparisons of outcomes between subjects correctly classified by the model with subjects misclassified suggested that the neural network may be able to identify latent predictive information from the MR images not incorporated in the ground truth labels. The residual learning model shows promise in the classification of MR images from subjects with TBI.

To investigate whether clinical decision support that automates the matching of ordered drugs to problems (clinical diagnoses) on the problem list can enhance the maintenance of both medication and problem lists in the electronic medical record, we designed a clinical decision support system to match ordered drugs on the medication list and ongoing problems on the problem list. We evaluated the capability and performance of this clinical decision support system in medication-problem matching using physician expert chart audits to match ordered drugs to ongoing clinical problems. A clinical decision support system was shown to be useful in improving medication-problem matches in 140 randomly selected audited patient encounters in three inpatient units. Enhanced maintenance of both the medication and problem lists can permit the exploitation of advanced decision support strategies that yield higher patient safety.

Preventable medical errors are a major problem in healthcare. Maintenance of accurate problem lists and drug lists in the electronic medical record is critical to the practice of medicine and patient safety. We built a simulator of a clinical decision support system that automates the process of maintaining an electronic problem list by processing drug order requests from a simulated computerized physician order entry system. Preliminary results revealed that the productivity of the system increased when the user interface was improved. This study highlights the importance of an enhanced user interface as a frontline in expediting physician data entry, streamlining better system workflow, and maintaining an electronic problem list effectively to reduce preventable medical errors and promote patient safety.

Downloading medical images on the Web creates certain compromises. The tradeoff is between higher resolution and faster download times. As resolution increases, download times increase. High-resolution (photographic quality) electronic images can potentially play a key role in medical education and patient care. On the Internet, images are typically formatted as Graphics Interchange Format (GIF) or the Joint Photographic Experts Group (JPEG) files. However, these formats are associated with considerable data loss in both color depth and image resolution. Furthermore, these images are available in a single resolution and have no capability of allowing the user to adjust resolution as needed. Images in the photo compact disc (PCD) format have higher resolutions than GIF or JPEG, but suffer the disadvantage of large file sizes leading to long download times on the Web. Furthermore, native web browsers are not currently able to read PCD files. The FlashPix format (FPX) offers distinct advantages over the PCD, GIF, and JPEG formats for display of high-resolution images on the Web. A Java applet can be easily downloaded for viewing FPX images. FPX images are higher resolution than JPEG and GIF images. FPX images offer rich resolutions comparable to PCD images with shorter download times.

Interactive laserdiscs are of limited value in large group learning situations due to the expense of establishing multiple workstations. The authors implemented an alternative to laserdisc video by using indexed digital video combined with an expert system. High-quality video was captured from a laserdisc player and combined with waveform audio into an audio-video-interleave (AVI) file format in the Microsoft Video-for-Windows environment (Microsoft Corp., Seattle, WA). With the use of an expert system, a knowledge-based computer program provided random access to these indexed AVI files. The program can be played on any multimedia computer without the need for laserdiscs. This system offers a high level of interactive video without the overhead and cost of a laserdisc player.

A treatment risk analysis system was developed and presented by the MEDAS project group. Now, a risk analysis system for medical tests has also been developed to form a complete medical decision support system. Moreover, the current system (a combined treatment and test system) has been redesigned and developed using an object-oriented hypermedia programming language, called Spinnaker Plus, running on the MS-Windows environment. The system has already been used to analyze several neurological problems.>

The ideal neurological consultation should be complete and legible. It should serve as a permanent and retrievable record. It should have a high clinical and educational value. The authors have built two prototypes of computer-based neurological consultation systems to meet these objectives. Both systems utilize methods of hypermedia and expert systems to enhance the quality of the computer-based neurological consultation.>

LITREF was developed to run on a microcomputer to support MAIESTRO, an expert system for the diagnosis and management of stroke cases. The architecture of LITREF uses an inverted file structure with a bitmap strategy. The evaluation process uses the cosine function to measure the similarity between a query and an abstract. The Salton interpolation process is used in the computation of recall and precision values. The experiment involves applying alternative suffixing algorithms to both index terms and query keywords. When the results are evaluated by comparing precision values at each recall level, it is found that the word-stem index method is superior to the full-word index method.>

A knowledge acquisition procedure is presented that is an alternative to the interview process, the traditional method used in knowledge engineering. In an effort to produce a knowledge base in a more timely and efficient manner, an automated procedure is developed. This process examines historical patient cases and generates expert system production rules from them. These rules form an initial knowledge base which can then be honed by the domain expert into the final knowledge base for an expert system. It is found that this automated procedure generates production rules that are equal in value to those produced through the interview process.>

An online bibliographic retrieval system, called the literature reference (Litref) system, is presented. It is designed to run on a high-end microcomputer to support the stroke consultant system (Maiestro). Litref interprets a Boolean command language derived from knowledge-index to provide fast, efficient browsing of abstracts from the stroke literature. User feedback on keywords derived from a relational thesaurus is used to enhance the original query. The system is designed to work in a microcomputer environment. Features and goals, design principles, and system architecture are discussed.>