IEEE Transactions on Biomedical Engineering最新文献

For people with Type 1 diabetes (T1D), accurate blood glucose (BG) forecasting is crucial for the effective delivery of insulin by Artificial Pancreas (AP) systems. Deep learning frameworks like Long Short-Term-Memory (LSTM) have been widely used to predict BG using continuous glucose monitor (CGM) data. However, these methods usually require large amounts of training data for personalized forecasts. Moreover, individuals with diabetes exhibit diverse glucose variability (GV), resulting in varying forecast accuracy. To address these limitations, we propose a novel deep learning framework: Incrementally Retrained Stacked LSTM (IS-LSTM). This approach gradually adapts to individuals' data and employs parameter-transfer for efficiency. We compare our method to three benchmarks using two CGM datasets from individuals with T1D: OpenAPS and Replace-BG. On both datasets, our approach significantly reduces root mean square error compared to the state of the art (Stacked LSTM): from 14.55 to 10.23mg/dL (OpenAPS) and 17.15 to 13.41mg/dL (Replace-BG) at 30-minute Prediction Horizon (PH). Clarke error grid analysis demonstrates clinical feasibility with at least 98.81% and 97.25% of predictions within the clinically safe zone at 30- and 60-minute PHs. Further, we demonstrate the effectiveness of our method in cold-start scenarios, which helps new CGM users obtain accurate predictions.

Objective: Accurate alignment of long bone fractures under minimally invasive procedures is a prerequisite for excellent treatment outcomes. However, the existing technologies suffer from the drawbacks of complex operations and excessive dependence on the surgeon's expertise. To solve these problems, we have developed a novel computer-assisted system to achieve rapid and effective reduction of fractures.

Methods: The automatic registration of the bone-fixator is accomplished based on the principal component analysis and the markers recognition. Then, the fracture reduction target is acquired by utilizing the Iterative Closest Point algorithm on the mirrored contralateral bone model. Next, the optimal reduction trajectory is automatically generated by considering collision detection, muscle pull force analysis, and trajectory optimization. Finally, the strut adjustment plan of the fixator is provided to the surgeon, combined with the results of bone-fixator registration.

Result: Modeling experiments verified the high accuracy of the system registration and the superiority of the reduction planning method, and clinical trials demonstrated the effectiveness and feasibility of the proposed system for fracture treatment.

Conclusion: The proposed system facilitates accurate and efficient planning of fracture reduction for surgeons through simple manipulation.

Significance: Our system enables a one-stop automatic acquisition of prescriptions for external fixation treatment of fractures.

Objective: The isolated mammalian retina may serve as a sensitive biosensor for preclinical drug testing, including eye drugs and a broader range of pharmaceuticals. To facilitate testing with minimal amounts of drug molecules or nanostructures, we developed a closed-perfusion transretinal electroretinography (tERG) setup.

Methods: The major challenge with small amounts of circulating perfusate was maintaining retinal viability and stability during long experiments. We conducted ex vivo tERG using WT C57BL/6J and Gnat1^-/- mice to assess rod- and cone-mediated light signals. The dark-adapted retina was stimulated with full-field light flashes while perfused at 5-6 ml/min.

Results: The minimum perfusate needed in our closed-circulation was around 50 ml. Penicillin-Streptomycin (Pen-Strep) was indispensable for long recordings. Rod responses remained stable for at least 42 hours, the longest recording we conducted, with the retina still responsive, and rod and cone bipolar cell responses for up to 12 hours. IBMX (3-isobutyl-1-methylxanthine), a non-specific phosphodiesterase (PDE) inhibitor with reversible effects, validated our setup. We used our setup to test the zwitterionic polymer poly(sulfobetaine methacrylate) (PSMBA), serving as a promising material for thermoresponsive nanostructures, and the corresponding monomer SBMA for possible harmful effects on mouse rod and bipolar cell functioning.

Conclusion: Our closed-perfusion tERG setup enables long experiments with small amounts of perfusate. PSMBA or SBMA had no effect on rod and bipolar cell responses.

Significance: This method is applicable for assessing drug functionality, as well as conducting preliminary biocompatibility and toxicity testing using small amounts of molecules or nanostructures that could impact neuronal or synaptic function.

Objective: A promising avenue for vision restoration against retinal degeneration is the use of semiconductor-based photovoltaic biointerfaces to substitute natural photoreceptors. Instead of silicon, perovskite has emerged as an exciting material for solar energy harvesting, and its nanocrystalline forms generally offer better stability than their bulk counterparts in addition to the distinct synthesis and fabrication steps.

Methods: Herein, we synthesize tetramethylammonium lead iodide (TMAPbI3) perovskite quantum dots (QDs) as a novel photoactive material for photovoltaic biointerfaces. While the TMAPbI3 quantum dots and electrolyte interface induces Faradaic photocurrent under light illumination, the heterojunction with P3HT converts the charge-transfer process to a safe capacitive photocurrent with an improved ionic responsivity of 17.4 mA/W.

Significance: The integration of the 18-nm quantum dot thickness shows good biocompatibility with primary cultures of hippocampal neurons and the photoresponse of the biointerface triggered photostimulation of the neurons. The rise of perovskite materials can stimulate novel forms of photovoltaic retina implants.