Predicting seizures ahead of time will have a significant positive clinical impact for people with epilepsy. Advances in machine learning/artificial intelligence (ML/AI) has provided us the tools needed to perform such predictive tasks. To date, advanced deep learning (DL) architectures such as the convolutional neural network (CNN) and long short-term memory (LSTM) have been used with mixed results. However, highly connected activity exhibited by epileptic seizures necessitates the design of more complex ML techniques which can better capture the complex interconnected neurological processes. Other challenges include the variability of EEG sensor data quality, different epilepsy and seizure profiles, lack of annotated datasets and absence of ML-ready benchmarks. In addition, successful models will need to perform inference in almost real-time using limited hardware compute-capacity. To address these challenges, we propose a lightweight architecture, called ESPFormer, whose novelty lies in the simple and smaller model-size and a lower computational load footprint needed to infer in real-time compared to other works in the literature. To quantify the performance of this lightweight model, we compared its performance with a custom-designed residual neural network (ResNet), a pre-trained vision transformer (ViT) and a pre-trained large-language model (LLM). We tested ESPFormer on MLSPred-Bench which is the largest patient-independent seizure prediction dataset comprising 12 benchmarks. Our results demonstrate that ESPFormer provides the best performance in terms of prediction accuracy for 4/12 benchmarks with an average improvement of 2.65% compared to the LLM, 3.35% compared to the ViT and 17.65% compared to the ResNet – and comparable results for other benchmarks. Our results indicate that lightweight transformer architecture may outperform resource-intensive LLM based models for real-time EEG-based seizure predictions.

Autism spectrum disorder (ASD) affects large number of children and adults in the US, and worldwide. Early and quick diagnosis of ASD can improve the quality of life significantly both for patients and their families. Prior research provides strong evidence that structural and functional magnetic resonance imaging (MRI) data collected from individuals with ASD exhibit distinguishing characteristics that differ in local and global, spatial and temporal neural patterns of the brain - and therefore can be used for diagnostic purposes for various mental disorders. However, the data from MRI are high-dimensional and advanced methods are needed to make sense out of these datasets. In this paper, we present a novel model based on graph convolutional network (GCN) that can utilize resting state fMRI (rs-fMRI) data to classify ASD subjects from health controls (HC). In addition to using the graph from traditional correlation matrices, our proposed GCN model incorporates graphlet topological counting as one of the training features. Our results show that graphlets can preserve the topological information of the graphs obtained from fMRI data. Combined with our GCN, the graphlets retain enough topological information to differentiate between the ASD and HC. Our proposed model gives an average accuracy of 64.27% on the whole ABIDE-I data sets (1035 subjects) and highest site-specific accuracy of 75.9%, which is comparable to other state-of-the-art methods - while potentially open to being more interpretable.

The Sparse Fast Fourier Transform (MIT-SFFT) is an algorithm to compute the discrete Fourier transform of a signal with a sublinear time complexity, i.e. algorithms with runtime complexity proportional to the sparsity level k, where k is the number of non-zero coefficients of the signal in the frequency domain. In this paper, we propose a highly scalable GPU-based parallel algorithm called GPU-SFFT for computing the SFFT of k-sparse signals. Our implementation of GPU-SFFT is based on parallel optimizations that leads to enormous speedups. These include carefully crafting parallel regions in the sequential MIT-SFFT code to exploit parallelism, and minimizing data movement between the CPU and the GPU. This allows us to exploit extreme parallelism for the CPU-GPU architectures and to maximize the number of concurrent threads executing instructions. Our experiments show that our designed CPU-GPU specific optimizations lead to enormous decrease in the run times needed for computing the SFFT. Further we show that GPU-SFFT is 38x times faster than the MIT-SFFT and 5x faster than cuFFT, the NVIDIA CUDA Fast Fourier Transform (FFT) library. The source code for GPU-SFFT is available at https://github.com/pcdslab.

In the field of data science, a huge amount of data, generally represented as graphs, needs to be processed and analyzed. It is of utmost importance that this data be processed swiftly and efficiently to save time and energy. The volume and velocity of data, along with irregular access patterns in graph data structures, pose challenges in terms of analysis and processing. Further, a big chunk of time and energy is spent on analyzing these graphs on large compute clusters and/or data-centers. Filtering and refining of data using graph sampling techniques are one of the most effective ways to speed up the analysis. Efficient accelerators, such as FPGAs, have proven to significantly lower the energy cost of running an algorithm. To this end, we present the design and implementation of a parallel graph sampling technique, for a large number of input graphs streaming into a FPGA. A parallel approach using OpenCL for FPGAs was adopted to come up with a solution that is both time- and energy-efficient. We introduce a novel graph data structure, suitable for streaming graphs on FPGAs, that allows time- and memory-efficient representation of graphs. Our experiments show that our proposed technique is 3x faster and 2x more energy efficient as compared to serial CPU version of the algorithm.

High-throughput mass spectrometers can produce thousands of peptide spectra from a single complex protein sample in a short amount of time. These data sets contain a substantial amount of redundancy (i.e. the same peptide is selected and identified multiple times in a single experiment) from peptides that may get selected multiple times in the liquid chromatography mass spectrometry (LC-MS/MS) experiment. The data from these mass spectrometers contain a substantial number of spectra that have low signal to noise (S/N) ratio and may not get interpreted due to poor quality. Recently, we presented a graph theoretic algorithm, CAMS (Clustering Algorithm for Mass Spectra) for clustering mass spectrometry data. CAMS utilized a novel metric, called a F-set, that allows accurate identification of the spectra that are similar with much higher accuracy and sensitivity than if single peak comparisons were performed. In this paper we present a multithreaded algorithm, called P-CAMS, for clustering of mass spectral data on multicore machines. The algorithm relies on intelligent matrix completion for graph construction and a load-balancing scheme for substantial speedups. We study the scalability performance of the proposed parallel algorithm on a multicore machine using synthetically generated spectra with parameters carefully chosen to mimic real-world mass spectrometry datasets. Real experimental datasets were also generated for quality assessment of the clustering results from the proposed algorithm. The results show that the proposed algorithms have scalable runtime performances and gives clustering results similar to a serial algorithm. The study also provides insight into the design of high performance algorithms for irregular problems in proteomics on many-core architectures.