Two-Dimensional Deep Learning Frameworks for Drug-Induced Cardiotoxicity Detection.

The identification of drug-induced cardiotoxicity remains a pressing challenge with far-reaching clinical and economic ramifications, often leading to patient harm and resource-intensive drug recalls. Current methodologies, including in vivo and in vitro models, have severe limitations in accurate identification of cardiotoxic substances. Pioneering a paradigm shift from these conventional techniques, our study presents two deep learning-based frameworks, STFT-CNN and SST-CNN, to assess cardiotoxicity with markedly improved accuracy and reliability. Leveraging the power of induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) as a more human-relevant cell model, we record mechanical beating signals through impedance measurements. These temporal signals were converted into enriched two-dimensional representations through advanced transformation techniques, specifically short-time Fourier transform (STFT) and synchro-squeezing transform (SST). These transformed data are fed into the proposed frameworks for comprehensive analysis, including drug type classification, concentration classification, cardiotoxicity classification, and new drug identification. Compared to traditional models like recurrent neural network (RNN) and 1-dimensional convolutional neural network (1D-CNN), SST-CNN delivered an impressive test accuracy of 98.55% in drug type classification and 99% in distinguishing cardiotoxic and noncardiotoxic drugs. Its feasibility is further highlighted with a stellar 98.5% average accuracy for classification of various concentrations, and the superiority of our proposed frameworks underscores their promise in revolutionizing drug safety assessments. With a potential for scalability, they represent a major leap in drug safety assessments, offering a pathway to more robust, efficient, and human-relevant cardiotoxicity evaluations.

Melatonin suppresses Akt/mTOR/S6K activity, induces cell apoptosis, and synergistically inhibits cell growth with sunitinib in renal carcinoma cells via reversing Warburg effect.

BACKGROUND: Metabolic alteration drives renal cell carcinoma (RCC) development, while the impact of melatonin (MLT), a neurohormone secreted during darkness, on RCC cell growth and underlying mechanisms remains unclear. METHODS: We detected concentration of metabolites through metabolomic analyses using UPLC-MS/MS, and the oxygen consumption rate was determined using the Seahorse Extracellular Flux analyzer. RESULTS: We observed that MLT effectively inhibited RCC cell growth both in vitro and in vivo. Additionally, MLT increased ROS levels, suppressed antioxidant enzyme activity, and induced apoptosis. Furthermore, MLT treatment upregulated key TCA cycle metabolites while reducing aerobic glycolysis products, leading to higher oxygen consumption rate, ATP production, and membrane potential. Moreover, MLT treatment suppressed phosphorylation of Akt, mTOR, and p70 S6 Kinase as well as the expression of HIF-1α/VEGFA in RCC cells; these effects were reversed by NAC (ROS inhibitors). Conversely, MLT synergistically inhibited cell growth with sunitinib and counteracted the Warburg effect induced by sunitinib in RCC cells. CONCLUSIONS: In conclusion, our results indicate that MLT treatment reverses the Warburg effect and promotes intracellular ROS production, which leads to the suppression of Akt/mTOR/S6K signaling pathway, induction of cell apoptosis, and synergistically inhibition of cell growth with sunitinib in RCC cells. Overall, this study provides new insights into the mechanisms underlying anti-tumor effect of MLT in RCC cells, and suggests that MLT might be a promising therapeutic for RCC.

Extracting Useful Emergency Information from Social Media: A Method Integrating Machine Learning and Rule-Based Classification.

User-generated contents (UGCs) on social media are a valuable source of emergency information (EI) that can facilitate emergency responses. However, the tremendous amount and heterogeneous quality of social media UGCs make it difficult to extract truly useful EI, especially using pure machine learning methods. Hence, this study proposes a machine learning and rule-based integration method (MRIM) and evaluates its EI classification performance and determinants. Through comparative experiments on microblog data about the "July 20 heavy rainstorm in Zhengzhou" posted on China's largest social media platform, we find that the MRIM performs better than pure machine learning methods and pure rule-based methods, and that its performance is influenced by microblog characteristics such as the number of words, exact address and contact information, and users' attention. This study demonstrates the feasibility of integrating machine learning and rule-based methods to mine the text of social media UGCs and provides actionable suggestions for emergency information management practitioners.

A biosensing system employing nonlinear dynamic analysis-assisted neural network for drug-induced cardiotoxicity assessment.

Preclinical investigation of drug-induced cardiotoxicity is of importance for drug development. To evaluate such cardiotoxicity, in vitro high-throughput interdigitated electrode-based recording of cardiomyocytes mechanical beating is widely used. To automatically analyze the features from the beating signals for drug-induced cardiotoxicity assessment, artificial neural network analysis is conventionally employed and signals are segmented into cycles and feature points are located in the cycles. However, signal segmentation and location of feature points for different signal shapes require design of specific algorithms. Consequently, this may lower the efficiency of research and the applications of such algorithms in signals with different morphologies are limited. Here, we present a biosensing system that employs nonlinear dynamic analysis-assisted neural network (NDANN) to avoid the signal segmentation process and directly extract features from beating signal time series. By processing beating time series with fixed time duration to avoid the signal segmentation process, this NDANN-based biosensing system can identify drug-induced cardiotoxicity with accuracy over 0.99. The individual drugs were classified with high accuracies over 0.94 and drug-induced cardiotoxicity levels were accurately predicted. We also evaluated the generalization performance of the NDANN-based biosensing system in assessing drug-induced cardiotoxicity through an independent dataset. This system achieved accuracy of 0.85-0.95 for different drug concentrations in identification of drug-induced cardiotoxicity. This result demonstrates that our NDANN-based biosensing system has the capacity of screening newly developed drugs, which is crucial in practical applications. This NDANN-based biosensing system can work as a new screening platform for drug-induced cardiotoxicity and improve the efficiency of bio-signal processing.

A multi-feature deep learning system to enhance glaucoma severity diagnosis with high accuracy and fast speed.

Glaucoma is the leading cause of irreversible blindness, and the early detection and timely treatment are essential for glaucoma management. However, due to the interindividual variability in the characteristics of glaucoma onset, a single feature is not yet sufficient for monitoring glaucoma progression in isolation. There is an urgent need to develop more comprehensive diagnostic methods with higher accuracy. In this study, we proposed a multi- feature deep learning (MFDL) system based on intraocular pressure (IOP), color fundus photograph (CFP) and visual field (VF) to classify the glaucoma into four severity levels. We designed a three-phase framework for glaucoma severity diagnosis from coarse to fine, which contains screening, detection and classification. We trained it on 6,131 samples from 3,324 patients and tested it on independent 240 samples from 185 patients. Our results show that MFDL achieved a higher accuracy of 0.842 (95 % CI, 0.795-0.888) than the direct four classification deep learning (DFC-DL, accuracy of 0.513 [0.449-0.576]), CFP-based single-feature deep learning (CFP-DL, accuracy of 0.483 [0.420-0.547]) and VF-based single-feature deep learning (VF-DL, accuracy of 0.725 [0.668-0.782]). Its performance was statistically significantly superior to that of 8 juniors. It also outperformed 3 seniors and 1 expert, and was comparable with 2 glaucoma experts (0.842 vs 0.854, p = 0.663; 0.842 vs 0.858, p = 0.580). With the assistance of MFDL, junior ophthalmologists achieved statistically significantly higher accuracy performance, with the increased accuracy ranged from 7.50 % to 17.9 %, and that of seniors and experts were 6.30 % to 7.50 % and 5.40 % to 7.50 %. The mean diagnosis time per patient of MFDL was 5.96 s. The proposed model can potentially assist ophthalmologists in efficient and accurate glaucoma diagnosis that could aid the clinical management of glaucoma.

A biosensing system employing nanowell microelectrode arrays to record the intracellular potential of a single cardiomyocyte.

Electrophysiological recording is a widely used method to investigate cardiovascular pathology, pharmacology and developmental biology. Microelectrode arrays record the electrical potential of cells in a minimally invasive and high-throughput way. However, commonly used microelectrode arrays primarily employ planar microelectrodes and cannot work in applications that require a recording of the intracellular action potential of a single cell. In this study, we proposed a novel measuring method that is able to record the intracellular action potential of a single cardiomyocyte by using a nanowell patterned microelectrode array (NWMEA). The NWMEA consists of five nanoscale wells at the center of each circular planar microelectrode. Biphasic pulse electroporation was applied to the NWMEA to penetrate the cardiomyocyte membrane, and the intracellular action potential was continuously recorded. The intracellular potential recording of cardiomyocytes by the NWMEA measured a potential signal with a higher quality (213.76 ± 25.85%), reduced noise root-mean-square (~33%), and higher signal-to-noise ratio (254.36 ± 12.61%) when compared to those of the extracellular recording. Compared to previously reported nanopillar microelectrodes, the NWMEA could ensure single cell electroporation and acquire high-quality action potential of cardiomyocytes with reduced fabrication processes. This NWMEA-based biosensing system is a promising tool to record the intracellular action potential of a single cell to broaden the usage of microelectrode arrays in electrophysiological investigation.

A biosensing system using a multiparameter nonlinear dynamic analysis of cardiomyocyte beating for drug-induced arrhythmia recognition.

Cardiovascular disease is the number one cause of death in humans. Therefore, cardiotoxicity is one of the most important adverse effects assessed by arrhythmia recognition in drug development. Recently, cell-based techniques developed for arrhythmia recognition primarily employ linear methods such as time-domain analysis that detect and compare individual waveforms and thus fall short in some applications that require automated and efficient arrhythmia recognition from large datasets. We carried out the first report to develop a biosensing system that integrated impedance measurement and multiparameter nonlinear dynamic algorithm (MNDA) analysis for drug-induced arrhythmia recognition and classification. The biosensing system cultured cardiomyocytes as physiologically relevant models, used interdigitated electrodes to detect the mechanical beating of the cardiomyocytes, and employed MNDA analysis to recognize drug-induced arrhythmia from the cardiomyocyte beating recording. The best performing MNDA parameter, approximate entropy, enabled the system to recognize the appearance of sertindole- and norepinephrine-induced arrhythmia in the recording. The MNDA reconstruction in phase space enabled the system to classify the different arrhythmias and quantify the severity of arrhythmia. This new biosensing system utilizing MNDA provides a promising and alternative method for drug-induced arrhythmia recognition and classification in cardiological and pharmaceutical applications.

Asp305Gly mutation improved the activity and stability of the styrene monooxygenase for efficient epoxide production in Pseudomonas putida KT2440.

BACKGROUND: Styrene monooxygenase (SMO) catalyzes the first step of aromatic alkene degradation yielding the corresponding epoxides. Because of its broad spectrum of substrates, the enzyme harbors a great potential for an application in medicine and chemical industries. RESULTS: In this study, we achieved higher enzymatic activity and better stability towards styrene by enlarging the ligand entrance tunnel and improving the hydrophobicity through error-prone PCR and site-saturation mutagenesis. It was found that Asp305 (D305) hindered the entrance of the FAD cofactor according to the model analysis. Therefore, substitution with amino acids possessing shorter side chains, like glycine, opened the entrance tunnel and resulted in up to 2.7 times higher activity compared to the wild-type enzyme. The half-lives of thermal inactivation for the variant D305G at 60 °C was 28.9 h compared to only 3.2 h of the wild type SMO. Moreover, overexpression of SMO in Pseudomonas putida KT2440 with NADH regeneration was carried out in order to improve biotransformation efficiency for epoxide production. A hexadecane/buffer (v/v) biphasic system was applied in order to minimize the inactivation effect of high substrate concentrations on the SMO enzyme. Finally, SMO activities of 190 U/g CDW were measured and a total amount of 20.5 mM (S)-styrene oxide were obtained after 8 h. CONCLUSIONS: This study offers an alternative strategy for improved SMO expression and provides an efficient biocatalytic system for epoxide production via engineering the entrance tunnel of the enzyme's active site.

Photonics-enabled compressive sensing with spectral encoding using an incoherent broadband source.

In this Letter, we propose an approach to achieving photonics-enabled compressive sensing of sparse wideband radio frequency signals in which an incoherent broadband source is applied, and the mixing and integration functions are realized in the optical domain. A spectrum shaper is employed to slice and encode the spectrum of the broadband light according to a predetermined random sequence. Because of the dispersion-induced group delay, the mixing between the incoming signal and the random bit sequence is achieved. At the output of the spectrum shaper, an array of photodetectors is employed to realize down-sampling, and the input sparse signal can be captured in a single-shot mode. Since no pulsed laser is employed, our scheme obviates the need for time-domain synchronization between the repetitive ultra-fast pulses and the random sequence. Experimental demonstrations and numerical results are presented to verify the feasibility and potential of the approach.

All-positive-coefficient microwave photonic filter with rectangular response.

In conventional approaches to realizing microwave photonic filters (MPFs) with rectangular response, both positive and negative coefficients should be included in the systems, which is a difficulty in incoherent MPFs. Electrical or optical ways to realize the MPFs with bipolar taps, such as those using balanced detectors or based on two opposite modulation slopes of dual-input Mach-Zehnder modulators, usually involve more components compared with the MPFs with unipolar taps. In this Letter, a design of MPFs with rectangular response using all positive coefficients is proposed. There is only one optical link as well as one photodetector utilized in the approach, which largely simplifies the filter structure. Experimental demonstrations of 55-positive-tap MPFs with different passbands are presented to verify the feasibility and potential of the approach.

Analysis of compressive sensing with optical mixing using a spatial light modulator.

Compressive sensing (CS) in a photonic link has a high potential for acquisition of wideband sparse signals. In CS it is necessary to mix the input sparse signal with a pseudorandom sequence prior to subsampling. A pulse shaper with a spatial light modulator (SLM) can be used in photonic CS as an optical mixer to improve the speed of mixing. In this approach, the sparse signal is modulated on a chirped optical pulse and the pseudorandom sequence is recorded on the SLM within the pulse shaper. The optical mixing in the frequency domain is realized based on the principle of frequency-to-time mapping. In this paper, we investigate the performance and limitations of photonic CS with an SLM in detail. A theoretical model to describe optical mixing based on frequency-to-time mapping is presented. We point out that there is an upper limit on the length of the pseudorandom sequence recorded on the SLM that can be mixed with the sparse signal due to the condition of the far-field approximation of the frequency-to-time mapping. Since the length of the pseudorandom sequence is one of the major factors that affect the signal recovery performance in CS, this limitation should be fully considered in the system design of the CS with optical mixing in the frequency domain. We present numerical and experimental results to verify the theoretical findings. Discussion on the performance improvement is also presented.