Alzheimer’s disease is the primary cause of dementia worldwide, with an increasing morbidity burden that may outstrip diagnosis and management capacity as the population ages. Current methods integrate patient history, neuropsychological testing and MRI to identify likely cases, yet effective practices remain variably applied and lacking in sensitivity and specificity. Here we report an interpretable deep learning strategy that delineates unique Alzheimer’s disease signatures from multimodal inputs of MRI, age, gender, and Mini-Mental State Examination score. Our framework linked a fully convolutional network, which constructs high resolution maps of disease probability from local brain structure to a multilayer perceptron and generates precise, intuitive visualization of individual Alzheimer’s disease risk en route to accurate diagnosis. The model was trained using clinically diagnosed Alzheimer’s disease and cognitively normal subjects from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset (n = 417) and validated on three independent cohorts: the Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing (AIBL) (n = 382), the Framingham Heart Study (n = 102), and the National Alzheimer’s Coordinating Center (NACC) (n = 582). Performance of the model that used the multimodal inputs was consistent across datasets, with mean area under curve values of 0.996, 0.974, 0.876 and 0.954 for the ADNI study, AIBL, Framingham Heart Study and NACC datasets, respectively. Moreover, our approach exceeded the diagnostic performance of a multi-institutional team of practicing neurologists (n = 11), and high-risk cerebral regions predicted by the model closely tracked post-mortem histopathological findings. This framework provides a clinically adaptable strategy for using routinely available imaging techniques such as MRI to generate nuanced neuroimaging signatures for Alzheimer’s disease diagnosis, as well as a generalizable approach for linking deep learning to pathophysiological processes in human disease.

IntroductionCoronavirus disease 2019 (COVID‐19) has rapidly become a global pandemic, but little is known about its potential impact on patients with myasthenia gravis (MG).MethodsWe studied the clinical course of COVID‐19 in five hospitalized patients with autoimmune MG (four with acetylcholine receptor antibodies, one with muscle‐specific tyrosine kinase antibodies) between April 1, 2020‐April 30‐2020.ResultsTwo patients required intubation for hypoxemic respiratory failure, whereas one required significant supplemental oxygen. One patient with previously stable MG had myasthenic exacerbation. One patient treated with tocilizumab for COVID‐19 was successfully extubated. Two patients were treated for MG with intravenous immunoglobulin without thromboembolic complications.DiscussionOur findings suggest that the clinical course and outcomes in patients with MG and COVID‐19 are highly variable. Further large studies are needed to define best practices and determinants of outcomes in this unique population.

Newborn infants must rapidly adjust their physiology and behavior to the specific demands of the novel postnatal environment. This adaptation depends, at least in part, on the infant's ability to learn from experiences. We report here that infants exhibit learning even while asleep. Bioelectrical activity from face and scalp electrodes was recorded from neonates during an eye movement conditioning procedure in which a tone was followed by a puff of air to the eye. Sleeping newborns rapidly learned the predictive relationship between the tone and the puff. Additionally, in the latter part of training, these infants exhibited a frontally maximum positive EEG slow wave possibly reflecting memory updating. As newborns spend most of their time sleeping, the ability to learn about external stimuli in the postnatal environment during nonawake states may be crucial for rapid adaptation and infant survival. Furthermore, because eyelid conditioning reflects functional cerebellar circuitry, this method potentially offers a unique approach for early identification of infants at risk for a range of developmental disorders including autism and dyslexia.EEG | eyelid conditioning | neonate D uring the first days of life, awake infants are capable of learning associations between oral motor patterns and altered milk flow (1) and can learn to alter sucking to produce a variety of reinforcers, including milk (2), their mother's voice (3, 4), or a sweet-tasting solution (5). Cross-sensory associative learning also has been demonstrated in awake neonates using paired auditory and visual stimuli (6, 7). Furthermore, awake newborns show Pavlovian conditioning to tactile (8) and taste stimuli (9, 10), as well as eyelid conditioning to paired auditory and tactile stimuli (11). These early adaptations to the postnatal environment have been well documented in awake infants, but as newborns spend the vast majority of their time asleep, the need and capacity to learn may not be confined to states of wakefulness.Even while asleep, neonates are able to process external information actively. Scalp recordings of brain activity in sleeping neonates have demonstrated their capacity to differentiate between two sounds (12-14), indicating that infants are forming representations of specific stimuli and distinguishing between those stimuli during sleep. We report here that sleeping neonates not only process information about individual events, but also learn about relationships between them.To investigate whether neonates can learn during sleep, we attempted to condition an eye movement response in 1-to 2-dayold infants while they slept. All infants were fed immediately before testing to increase the likelihood they would sleep through the entire procedure. Sleep status was confirmed using behavioral observations in conjunction with heart rate variability patterns, respiratory regularity, and video scoring of the infants' faces. Infants were videotaped while exposed to tones and puffs of air directed at the eyelid. In the experimental group, tones w...

Worldwide, there are nearly 10 million new cases of dementia annually, of which Alzheimer’s disease (AD) is the most common. New measures are needed to improve the diagnosis of individuals with cognitive impairment due to various etiologies. Here, we report a deep learning framework that accomplishes multiple diagnostic steps in successive fashion to identify persons with normal cognition (NC), mild cognitive impairment (MCI), AD, and non-AD dementias (nADD). We demonstrate a range of models capable of accepting flexible combinations of routinely collected clinical information, including demographics, medical history, neuropsychological testing, neuroimaging, and functional assessments. We then show that these frameworks compare favorably with the diagnostic accuracy of practicing neurologists and neuroradiologists. Lastly, we apply interpretability methods in computer vision to show that disease-specific patterns detected by our models track distinct patterns of degenerative changes throughout the brain and correspond closely with the presence of neuropathological lesions on autopsy. Our work demonstrates methodologies for validating computational predictions with established standards of medical diagnosis.

Peripheral neuropathies are a group of disorders that affect the peripheral nervous system, for which hundreds of etiologies have been identified. This article presents a stepwise approach to the evaluation and workup of peripheral neuropathy, which starts with a detailed history of symptoms, family and occupational history, and a neurological as well as general physical exam. Pattern recognition of various neuropathies can help to build a differential diagnosis based on the presentation. Such patterns include acute versus chronic, primary demyelinating versus axonal, hereditary versus acquired, asymmetric versus symmetric, presence of facial palsies, sensory or motor predominant, and presence of prominent autonomic symptoms. Early categorization of the type of neuropathy can help focus the workup for peripheral neuropathy. Nerve conduction studies and electromyography (NCS/EMG) is the primary diagnostic tool in the evaluation of patients with large-fiber polyneuropathy. One of the most important roles of NCS/EMG is to help categorize polyneuropathy as primary axonal versus primary demyelinating. The finding of a primary demyelinating polyneuropathy narrows the differential diagnosis of polyneuropathy dramatically and increases the chances of finding a treatable etiology. Laboratory workup includes serum studies and potentially cerebrospinal fluid, genetic studies, immunological markers, and fat pad biopsy for select patients. Skin biopsy may be used to assess intraepidermal nerve fiber density if small-fiber neuropathy is suspected, and nerve biopsy may be useful in select cases. In recent years, magnetic resonance imaging and neuromuscular ultrasound have also shown promise in the evaluation of peripheral neuropathy. Identification of the etiology of neuropathy is crucial and often time-sensitive, as an increasing number of causes are now reversible or treatable.

Peripheral neuropathy occurs in the setting of both hereditary and acquired amyloidosis. The most common form of hereditary amyloidosis is caused by 1 of 140 mutations in the transthyretin (TTR) gene, which can lead to neuropathic hereditary transthyretin amyloidosis (hATTR; previously referred to as transthyretin familial amyloid polyneuropathy), whereas acquired immunoglobulin light chain (AL) amyloidosis is the most common acquired form. Patients typically present with a sensorimotor polyneuropathy, focal neuropathy such as carpal tunnel syndrome, or autonomic neuropathy. When neuropathy is the sole or dominant presenting symptom, the diagnosis is commonly delayed. With the advent of new drug therapies for AL amyloidosis and hATTR amyloidosis, including proteasome inhibitors, TTR silencers, and TTR protein stabilizers, the neurologist is uniquely positioned to diagnose neurologic manifestations of systemic amyloidosis, leading to earlier disease identification and treatment. This article reviews the epidemiology, clinical presentations, pathophysiology, diagnostic workup, and treatment of neuropathy in the setting of amyloidosis.

Distal symmetric polyneuropathy (DSPN), the most common form of diabetic neuropathy, has a complex pathophysiology and can be a major source of physical and psychologic disability. The management of DSPN can be frustrating for both patient and physician. This article provides a general overview of typical patient pathways in DSPN, and highlights variations in diagnosis, management, and referral patterns among different providers. DSPN is managed in several settings by primary care physicians (PCPs), specialists, and nurse practitioners. The initial clinical management of the patient is often dependent on the presenting complaint, the referral pattern of the provider, level of comfort of the PCP in managing diabetic complications, and geographic access to specialists. The primary treatment of DSPN focuses mainly on glycemic control and adjustment of modifiable risk factors, but other causes of neuropathy should also be investigated. Several pharmacologic agents are recommended by treatment guidelines, and as DSPN typically exists with comorbid conditions, a multimodal therapeutic approach should be considered. Barriers to effective management include failure to recognize DSPN, and misdiagnosis. Patient education also remains important. Referral patterns vary widely according to geographic location, access to services, provider preferences, and comfort in managing complex aspects of the disease. The variability in patient pathways affects patient education, satisfaction, and outcomes. Standardized screening tools, a multidisciplinary team approach, and treatment algorithms for diabetic neuropathy should improve future care. To improve patient outcomes, DSPN needs to be diagnosed sooner and interventions made before significant nerve damage occurs.

The prevalence of those affected by HIV-DSP will continue to grow with the aging population of HIV-infected individuals. Compared to the diabetic neuropathy drug trials, trials in both symptomatic and disease-modifying agents for HIV-DSP have had little success. Other forms of HIV-related peripheral neuropathies are discussed briefly, and include acute and chronic inflammatory demyelinating polyneuropathy, autonomic neuropathy, polyradiculopathy, mononeuropathies, mononeuritis multiplex, cranial neuropathies, and amyotrophic lateral sclerosis-like motor neuropathy.