SERS Nanowire Chip and Machine Learning-Enabled Classification of Wild-Type and Antibiotic-Resistant Bacteria at Species and Strain Levels.

Antimicrobial resistance (AMR) is a major health threat worldwide and the culture-based bacterial detection methods are slow. Surface-enhanced Raman spectroscopy (SERS) can be used to identify target analytes in real time with sensitivity down to the single-molecule level, providing a promising solution for the culture-free bacterial detection. We report the fabrication of SERS substrates having tightly packed silver (Ag) nanoparticles loaded onto long silicon nanowires (Si NWs) grown by the metal-assisted chemical etching (MACE) method for the detection of bacteria. The optimized SERS chips exhibited sensitivity down to 10-12 M concentration of R6G molecules and detected reproducible Raman spectra of bacteria down to a concentration of 100 colony forming units (CFU)/mL, which is a thousand times lower than the clinical threshold of bacterial infections like UTI (105 CFU/mL). A Siamese neural network model was used to classify SERS spectra from bacteria specimens. The trained model identified 12 different bacterial species, including those which are causative agents for tuberculosis and urinary tract infection (UTI). Next, the SERS chips and another Siamese neural network model were used to differentiate AMR strains from susceptible strains of Escherichia coli (E. coli). The enhancement offered by SERS chip-enabled acquisitions of Raman spectra of bacteria directly in the synthetic urine by spiking the sample with only 103 CFU/mL E. coli. Thus, the present study lays the ground for the identification and quantification of bacteria on SERS chips, thereby offering a potential future use for rapid, reproducible, label-free, and low limit detection of clinical pathogens.

Multi-label classification of reduced-lead ECGs using an interpretable deep convolutional neural network.

Objective.We propose a model that can perform multi-label classification on 26 cardiac abnormalities from reduced lead Electrocardiograms (ECGs) and interpret the model.Approach.PhysioNet/computing in cardiology (CinC) challenge 2021 datasets are used to train the model. All recordings shorter than 20 s are preprocessed by normalizing, resampling, and zero-padding. The frequency domains of the recordings are obtained by applying fast Fourier transform. The time domain and frequency domain of the signals are fed into two separate deep convolutional neural networks. The outputs of these networks are then concatenated and passed through a fully connected layer that outputs the probabilities of 26 classes. Data imbalance is addressed by using a threshold of 0.13 to the sigmoid output. The 2-lead model is tested under noise contamination based on the quality of the signal and interpreted using SHapley Additive exPlanations (SHAP).Main results.The proposed method obtained a challenge score of 0.55, 0.51, 0.56, 0.55, and 0.56, ranking 2nd, 5th, 3rd, 3rd, and 3rd out of 39 officially ranked teams on 12-lead, 6-lead, 4-lead, 3-lead, and 2-lead hidden test datasets, respectively, in the PhysioNet/CinC challenge 2021. The model performs well under noise contamination with meanF1 scores of 0.53, 0.56 and 0.56 for the excellent, barely acceptable and unacceptable signals respectively. Analysis of the SHAP values of the 2-lead model verifies the performance of the model while providing insight into labeling inconsistencies and reasons for the poor performance of the model in some classes.Significance.We have proposed a model that can accurately identify 26 cardiac abnormalities using reduced lead ECGs that performs comparably with 12-lead ECGs and interpreted the model behavior. We demonstrate that the proposed model using only the limb leads performs with accuracy comparable to that using all 12 leads.