An artificial intelligence that increases simulated brain-computer interface performance.

Objective.Brain-computer interfaces (BCIs) translate neural activity into control signals for assistive devices in order to help people with motor disabilities communicate effectively. In this work, we introduce a new BCI architecture that improves control of a BCI computer cursor to type on a virtual keyboard.Approach.Our BCI architecture incorporates an external artificial intelligence (AI) that beneficially augments the movement trajectories of the BCI. This AI-BCI leverages past user actions, at both long (100 s of seconds ago) and short (100 s of milliseconds ago) timescales, to modify the BCI's trajectories.Main results.We tested our AI-BCI in a closed-loop BCI simulator with nine human subjects performing a typing task. We demonstrate that our AI-BCI achieves: (1) categorically higher information communication rates, (2) quicker ballistic movements between targets, (3) improved precision control to 'dial in' on targets, and (4) more efficient movement trajectories. We further show that our AI-BCI increases performance across a wide control quality spectrum from poor to proficient control.Significance.This AI-BCI architecture, by increasing BCI performance across all key metrics evaluated, may increase the clinical viability of BCI systems.

An integrated space-to-ground quantum communication network over 4,600 kilometres.

Quantum key distribution (QKD)1,2 has the potential to enable secure communication and information transfer3. In the laboratory, the feasibility of point-to-point QKD is evident from the early proof-of-concept demonstration in the laboratory over 32 centimetres4; this distance was later extended to the 100-kilometre scale5,6 with decoy-state QKD and more recently to the 500-kilometre scale7-10 with measurement-device-independent QKD. Several small-scale QKD networks have also been tested outside the laboratory11-14. However, a global QKD network requires a practically (not just theoretically) secure and reliable QKD network that can be used by a large number of users distributed over a wide area15. Quantum repeaters16,17 could in principle provide a viable option for such a global network, but they cannot be deployed using current technology18. Here we demonstrate an integrated space-to-ground quantum communication network that combines a large-scale fibre network of more than 700 fibre QKD links and two high-speed satellite-to-ground free-space QKD links. Using a trusted relay structure, the fibre network on the ground covers more than 2,000 kilometres, provides practical security against the imperfections of realistic devices, and maintains long-term reliability and stability. The satellite-to-ground QKD achieves an average secret-key rate of 47.8 kilobits per second for a typical satellite pass-more than 40 times higher than achieved previously. Moreover, its channel loss is comparable to that between a geostationary satellite and the ground, making the construction of more versatile and ultralong quantum links via geosynchronous satellites feasible. Finally, by integrating the fibre and free-space QKD links, the QKD network is extended to a remote node more than 2,600 kilometres away, enabling any user in the network to communicate with any other, up to a total distance of 4,600 kilometres.

Deep Learning Neural Encoders for Motor Cortex.

Intracortical brain-machine interfaces (BMIs) transform neural activity into control signals to drive a prosthesis or communication device, such as a robotic arm or computer cursor. To be clinically viable, BMI decoders must achieve high accuracy and robustness. Optimizing these decoders is expensive, traditionally requiring animal or human experiments spanning months to years. This is because BMIs are closed-loop systems, where the user updates his or her motor commands in response to an imperfectly decoded output. Decoder optimization using previously collected "offline" data will therefore not capture this closed-loop response. An alternative approach to significantly accelerate decoder optimization is to use a closed-loop experimental simulator. A key component of this simulator is the neural encoder, which synthetically generates neural population activity from kinematics. Prior neural encoders do not model important features of neural population activity. To overcome these limitations, we use deep learning neural encoders. We find these models significantly outperform prior neural encoders in reproducing peri-stimulus time histograms (PSTHs) and neural population dynamics. We also find that deep learning neural encoders better match neural decoding results in offline data and closed-loop experimental data. We anticipate these deep-learning neural encoders will substantially improve simulators for BMIs, enabling faster evaluation, optimization, and characterization of BMI decoder algorithms.

Clustering of unconventional cardiovascular risk factors among Taiwanese hemodialysis patients.

BACKGROUND: Cardiovascular disease (CAVD) is the most common cause of mortality for chronic hemodialysis (HD) patients, yet the risk factors for the events have not been well established. METHODS: We conducted a multicenter cross-sectional survey in 995 chronic HD patients recruited from 12 HD centers in Taiwan to investigate the prevalence of CAVD, including coronary heart disease (CHD), cerebrovascular disease (CVD), and peripheral vascular disease (PVD), and related them to 30 different parameters. RESULTS: The mean age of 995 patients (499 males/496 females) was 56.4 +/- 12.3 years, and average HD duration was 59.8 +/- 51.2 months. The prevalence rates of CHD, CVD, and PVD were 24.0, 6.0, and 5.3%, respectively. Results of our multivariate logistic regression analysis showed that out of the conventional CAVD risk factors, only old age and diabetes could be significantly associated with CAVD. Meanwhile, we found some novel clinical correlates, including low apolipoprotein A-I and creatinine for CHD, low uric acid for CVD, and low hematocrit and low diastolic blood pressure for PVD. Interestingly, left ventricular hypertrophy was found to be an independent correlate for all three: CHD, CVD, and PVD. CONCLUSIONS: Our study suggests that consideration of conventional cardiovascular risk factors as well as unconventional risk factors might better assess the risk for CAVD among HD patients.

Intrahippocampal administration of A beta(1-40) impairs spatial learning and memory in hyperglycemic mice.

Age-related neurodegenerative dementia, particularly Alzheimer's disease (AD), may be exacerbated by several interacting risk factors including genetic predisposition, beta amyloid (A beta) protein accumulation, environmental toxins, head trauma, and abnormal glycolytic metabolism. We examined the spatial learning and memory effects of A beta(1-40) administration on hyperglycemic mice by their performance in the Morris water maze. Chronic hyperglycemia was induced in male C57BL/6J mice to mimic diabetes mellitus by intraperitoneal injection of streptozotocin (STZ), which specifically destroys pancreatic beta-islet cells. Ten days after STZ treatment, intrahippocampal infusion of vehicle, monomer, or oligomer A beta(1-40) was given to these hyperglycemic mice. Our results demonstrate that in comparison with vehicle or monomer A beta(1-40), oligomer A beta(1-40) induced significant deficits of spatial learning and memory in hyperglycemic mice. Apoptotic signals were identified in the CA1 and dentate gyrus of hippocampus in hyperglycemic mice. A beta accumulation, oxidative stress, and apoptosis in the CA1 region were more intensive in hyperglycemic mice than that in normoglycemic mice after acute treatment with oligomer A beta(1-40) peptide treatment. These results indicate that CA1 apoptosis was enhanced by oxidative stress resulting from accumulation of A beta. Considered together, these findings suggest that hyperglycemic mice are more vulnerable to the A beta-induced-oxidative stress than normal subjects. We therefore propose that A beta accumulation would be enhanced by hyperglycemia, and the oxidative stress caused by A beta accumulation would in turn enhance the AD symptoms.