We present FlexiFlow, a lifetime-aware design framework for item-level intelligence (ILI) where computation is integrated directly into disposable products like food packaging and medical patches. Our framework leverages natively flexible electronics which offer significantly lower costs than silicon but are limited to kHz speeds and several thousands of gates. Our insight is that unlike traditional computing with more uniform deployment patterns, ILI applications exhibit 1000× variation in operational lifetime, fundamentally changing optimal architectural design decisions when considering trillion-item deployment scales. To enable holistic design and optimization, we model the trade-offs between embodied carbon footprint and operational carbon footprint based on application-specific lifetimes. The framework includes: (1) FlexiBench, a workload suite targeting sustainability applications from spoilage detection to health monitoring; (2) FlexiBits, area-optimized RISC-V cores with 1/4/8-bit datapaths achieving 2.65× to 3.50× better energy efficiency per workload execution; and (3) a carbon-aware model that selects optimal architectures based on deployment characteristics. We show that lifetime-aware microarchitectural design can reduce carbon footprint by 1.62×, while algorithmic decisions can reduce carbon footprint by 14.5×. We validate our approach through the first tape-out using a PDK for flexible electronics with fully open-source tools, achieving 30.9\,kHz operation. FlexiFlow enables exploration of computing at the Extreme Edge where conventional design methodologies must be reevaluated to account for new constraints and considerations. FlexiFlow is available at https://github.com/harvard-edge/FlexiFlow.

Medical vision-language models often struggle with generating accurate quantitative measurements in radiology reports, leading to hallucinations that undermine clinical reliability. We introduce FactCheXcker, a modular framework that de-hallucinates radiology report measurements by leveraging an improved query-code-update paradigm. Specifically, FactCheXcker employs specialized modules and the code generation capabilities of large language models to solve measurement queries generated based on the original report. After extracting measurable findings, the results are incorporated into an updated report. We evaluate FactCheXcker on endotracheal tube placement, which accounts for an average of 78% of report measurements, using the MIMIC-CXR dataset and 11 medical reportgeneration models. Our results show that FactCheXcker significantly reduces hallucinations, improves measurement precision, and maintains the quality of the original reports. Specifically, FactCheXcker improves the performance of all 11 models and achieves an average improvement of 135.0% in reducing measurement hallucinations measured by mean absolute error. Code is available at https://github.com/rajpurkarlab/FactCheXcker.

Periodontal disease is common among older adults, with around 50% of all adults aged 30 years and older having some form of gum disease. Although the disease might seem minor, without early or proper diagnosis, the disease can affect all supporting structures of the teeth, including gums, periodontal ligament, and alveolar bone. This research can help find patterns common among people with periodontal disease and possibly prevent the effects of it in the elderly. Here, we explored the deep learning capabilities of convolutional neural networks (CNN) and VGG16 models for detecting disease-associated patterns derived from 16S rRNA gene sequences of plaque samples. The CNN model achieved a 90.2% accuracy for differentiating periodontal disease from health samples, with 76% precision and 86% recall. This model could lead to earlier detection and more effective treatment of periodontal disease, ultimately improving patients’ dental health outcomes and quality of life.

Frequency Chaos Game Representation (FCGR) is a method where DNA sequences are converted into a 2D image representing k-mer nucleotide frequencies with additional geometric fractal patterns. We used FCGR to analyze nextgeneration sequences (NGS) of 16S rRNA gene amplicons derived from oral samples with or without caries in combination with machine deep learning to differentiate disease from health. The NGS data captured 16S rRNA genes of all species in the oral cavity for studying oral microbiome, which has been shown to associate with the progression of caries. Our hypothesis is that FCGR patterns of caries and caries-free samples are unique and can be differentiated by deep learning using convolutional neural network model (CNN) - a model specialized in classifying images. Samples labeled as caries and caries-free were split into 80-10-10 for training, testing, and validation purposes in CNN training. An accuracy of 87.5% was achieved in this pilot study. Our study showed that FCGR transforms DNA sequences into fractal image patterns and can be applied to oral microbiome NGS data for classifying clinical oral samples as caries or caries-free. This approach allows for fast and early diagnosis and prevention of major cavities.

We propose the Multimodal Clinical Benchmark for Emergency Care (MC-BEC), a comprehensive benchmark for evaluating foundation models in Emergency Medicine using a dataset of 100K+ continuously monitored Emergency Department visits from 2020-2022. MC-BEC focuses on clinically relevant prediction tasks at timescales from minutes to days, including predicting patient decompensation, disposition, and emergency department (ED) revisit, and includes a standardized evaluation framework with train-test splits and evaluation metrics. The multimodal dataset includes a wide range of detailed clinical data, including triage information, prior diagnoses and medications, continuously measured vital signs, electrocardiogram and photoplethysmograph waveforms, orders placed and medications administered throughout the visit, free-text reports of imaging studies, and information on ED diagnosis, disposition, and subsequent revisits. We provide performance baselines for each prediction task to enable the evaluation of multimodal, multitask models. We believe that MC-BEC will encourage researchers to develop more effective, generalizable, and accessible foundation models for multimodal clinical data.

Acute lymphoblastic leukemia (ALL) is one of the most common types of cancer among children. It can rapidly become fatal within weeks, and as such, early diagnosis is critical. Problematically, ALL diagnosis mainly involves manual blood smear analysis relying on the expertise of medical professionals, which is error-prone and time-consuming. Thus, it is important to develop artificial intelligence tools that will identify leukemic cells from a microscopic image faster, more accurately, and cheaper. Here, we investigate the capabilities of a traditional convolutional neural network(CNN) and You Only Look Once (YOLO) models for real-time detection of leukemic cells. The YOLOv5s model shows 97.2% accuracy for the task of object detection of ALL cells, with the inference speed allowing 80 image frames to be processed per second. These new findings can provide valuable insight in applying real-time object detection algorithms for improving the efficiency of blood cancer diagnosis.

OBJECTIVE: Machine learning (ML) models for allocating readmission-mitigating interventions are typically selected according to their discriminative ability, which may not necessarily translate into utility in allocation of resources. Our objective was to determine whether ML models for allocating readmission-mitigating interventions have different usefulness based on their overall utility and discriminative ability. MATERIALS AND METHODS: We conducted a retrospective utility analysis of ML models using claims data acquired from the Optum Clinformatics Data Mart, including 513,495 commercially-insured inpatients (mean [SD] age 69 [19] years; 294,895 [57%] Female) over the period January 2016 through January 2017 from all 50 states with mean 90 day cost of $11,552. Utility analysis estimates the cost, in dollars, of allocating interventions for lowering readmission risk based on the reduction in the 90-day cost. RESULTS: Allocating readmission-mitigating interventions based on a GBDT model trained to predict readmissions achieved an estimated utility gain of $104 per patient, and an AUC of 0.76 (95% CI 0.76, 0.77); allocating interventions based on a model trained to predict cost as a proxy achieved a higher utility of $175.94 per patient, and an AUC of 0.62 (95% CI 0.61, 0.62). A hybrid model combining both intervention strategies is comparable with the best models on either metric. Estimated utility varies by intervention cost and efficacy, with each model performing the best under different intervention settings. CONCLUSION: We demonstrate that machine learning models may be ranked differently based on overall utility and discriminative ability. Machine learning models for allocation of limited health resources should consider directly optimizing for utility.