Injectable ultrasonic sensor for wireless monitoring of intracranial signals.

Direct and precise monitoring of intracranial physiology holds immense importance in delineating injuries, prognostication and averting disease1. Wired clinical instruments that use percutaneous leads are accurate but are susceptible to infection, patient mobility constraints and potential surgical complications during removal2. Wireless implantable devices provide greater operational freedom but include issues such as limited detection range, poor degradation and difficulty in size reduction in the human body3. Here we present an injectable, bioresorbable and wireless metastructured hydrogel (metagel) sensor for ultrasonic monitoring of intracranial signals. The metagel sensors are cubes 2 × 2 × 2 mm3 in size that encompass both biodegradable and stimulus-responsive hydrogels and periodically aligned air columns with a specific acoustic reflection spectrum. Implanted into intracranial space with a puncture needle, the metagel deforms in response to physiological environmental changes, causing peak frequency shifts of reflected ultrasound waves that can be wirelessly measured by an external ultrasound probe. The metagel sensor can independently detect intracranial pressure, temperature, pH and flow rate, realize a detection depth of 10 cm and almost fully degrade within 18 weeks. Animal experiments on rats and pigs indicate promising multiparametric sensing performances on a par with conventional non-resorbable wired clinical benchmarks.

Inverse-Designed Aid Lenses for Precise Correction of Color Vision Deficiency.

Color vision deficiency (CVD) is a common ocular disorder affecting more than 300 million people on the earth. Although no clinical cure for the disorder currently exists, some specialized color filtering glasses/lenses based on dyes, metasurfaces, or nanocomposites have been employed for CVD management. However, as CVD patients usually diversify in their classification and severity, none of the current lenses provides a customized correction for various CVD patients, resulting in undesirable correction effects. Here, we present an inverse-designed approach for the precise correction of CVD. The wavelength shift of a patient's abnormal cone photoreceptors was measured to inversely design the best blocking wavelength and blocking rate of the lens. Then the customized aid lenses were fabricated using silica-coated gold nanoparticles with appropriate sizes and concentrations, verified by the simulated color vision and human tests. This study demonstrates the potential of the inverse-designed aid lenses in precise color filtering and customized CVD management.

A biological network-based regularized artificial neural network model for robust phenotype prediction from gene expression data.

BACKGROUND: Stratification of patient subpopulations that respond favorably to treatment or experience and adverse reaction is an essential step toward development of new personalized therapies and diagnostics. It is currently feasible to generate omic-scale biological measurements for all patients in a study, providing an opportunity for machine learning models to identify molecular markers for disease diagnosis and progression. However, the high variability of genetic background in human populations hampers the reproducibility of omic-scale markers. In this paper, we develop a biological network-based regularized artificial neural network model for prediction of phenotype from transcriptomic measurements in clinical trials. To improve model sparsity and the overall reproducibility of the model, we incorporate regularization for simultaneous shrinkage of gene sets based on active upstream regulatory mechanisms into the model. RESULTS: We benchmark our method against various regression, support vector machines and artificial neural network models and demonstrate the ability of our method in predicting the clinical outcomes using clinical trial data on acute rejection in kidney transplantation and response to Infliximab in ulcerative colitis. We show that integration of prior biological knowledge into the classification as developed in this paper, significantly improves the robustness and generalizability of predictions to independent datasets. We provide a Java code of our algorithm along with a parsed version of the STRING DB database. CONCLUSION: In summary, we present a method for prediction of clinical phenotypes using baseline genome-wide expression data that makes use of prior biological knowledge on gene-regulatory interactions in order to increase robustness and reproducibility of omic-scale markers. The integrated group-wise regularization methods increases the interpretability of biological signatures and gives stable performance estimates across independent test sets.

CRESPR: Modular sparsification of DNNs to improve pruning performance and model interpretability.

Modern DNNs often include a huge number of parameters that are expensive for both computation and memory. Pruning can significantly reduce model complexity and lessen resource demands, and less complex models can also be easier to explain and interpret. In this paper, we propose a novel pruning algorithm, Cluster-Restricted Extreme Sparsity Pruning of Redundancy (CRESPR), to prune a neural network into modular units and achieve better pruning efficiency. With the Hessian matrix, we provide an analytic explanation of why modular structures in a sparse DNN can better maintain performance, especially at an extreme high pruning ratio. In CRESPR, each modular unit contains mostly internal connections, which clearly shows how subgroups of input features are processed through a DNN and eventually contribute to classification decisions. Such process-level revealing of internal working mechanisms undoubtedly leads to better interpretability of a black-box DNN model. Extensive experiments were conducted with multiple DNN architectures and datasets, and CRESPR achieves higher pruning performance than current state-of-the-art methods at high and extremely high pruning ratios. Additionally, we show how CRESPR improves model interpretability through a concrete example.

Bioinspired Soft Elastic Metamaterials for Reconstruction of Natural Hearing.

Natural hearing which means hearing naturally like normal people is critical for patients with hearing loss to participate in life. Cochlear implants have enabled numerous severe hearing loss patients to hear voice functionally, while cochlear implant users can hardly distinguish different tones or appreciate music subject to the absence of rate coding and insufficient frequency channels. Here a bioinspired soft elastic metamaterial that reproduces the shape and key functions of the human cochlea is reported. Inspired by human cochlea, the metamaterials are designed to possess graded microstructures with high effective refractive index distributed on a spiral shape to implement position-related frequency demultiplexing, passive sound enhancements of 10 times, and high-speed parallel processing of 168-channel sound/piezoelectric signals. Besides, it is demonstrated that natural hearing artificial cochlea has fine frequency resolution up to 30 Hz, a wide audible range from 150-12 000 Hz, and a considerable output voltage that can activate the auditory pathway in mice. This work blazes a promising trail for reconstruction of natural hearing in patients with severe hearing loss.

Overcoming drug resistance with functional mesoporous titanium dioxide nanoparticles combining targeting, drug delivery and photodynamic therapy.

The resistance of tumor cells is a major cause of chemotherapy failure in cancer patients. Photodynamic therapy (PDT) as a noninvasive treatment strategy with high specificity is a promising method for the treatment of cancer. In this study, a CD44 and N-cadherin dual targeting drug delivery system in combination with mesoporous titanium dioxide nanoparticle (MTN)-based PDT has been successfully constructed for overcoming drug resistance. Hyaluronic acid (HA) and ADH-1 (a cyclic pentapeptide) were grafted onto the surface of MTN to construct ADH-1-HA-MTN, and doxorubicin (DOX) was selected as a model drug. HA can both trap DOX in the wells of MTN and target CD44-overexpressing tumor cells. ADH-1 blocks the EMT process of tumor cells by selectively inhibiting the function of N-cadherin. Besides, a large number of reactive oxygen species (ROS) were generated by MTN under X-ray irradiation, which could provide a cancer cell killing effect. Cytotoxicity tests showed that ADH-1-HA-MTN/DOX was more toxic to tumor cells than its non-ADH-1 modified counterparts. Western blotting analysis showed that ADH-1-HA-MTN/DOX overcame the drug resistance of tumor cells by preventing the process of epithelial-mesenchymal transition. Taken together, ADH-1-HA-MTN may be a promising targeted drug delivery system to overcome the drug resistance of tumors.