Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

XI. STRATEGIES FOR IMPROVING DIABETES CARE D iabetes is a chronic illness that requires continuing medical care and patient self-management education to prevent acute complications and to reduce the risk of long-term complications. Diabetes care is complex and requires that many issues, beyond glycemic control, be addressed. A large body of evidence exists that supports a range of interventions to improve diabetes outcomes. These standards of care are intended to provide clinicians, patients, researchers, payors, and other interested individuals with the components of diabetes care, treatment goals, and tools to evaluate the quality of care. While individual preferences, comorbidities, and other patient factors may require modification of goals, targets that are desirable for most patients with diabetes are provided. These standards are not intended to preclude more extensive evaluation and management of the patient by other specialists as needed. For more detailed information, refer to Bode (Ed.): Medical Management of Type 1 Diabetes (1), Burant (Ed): Medical Management of Type 2 Diabetes (2), and Klingensmith (Ed): Intensive Diabetes Management (3). The recommendations included are diagnostic and therapeutic actions that are known or believed to favorably affect health outcomes of patients with diabetes. A grading system (Table 1), developed by the American Diabetes Association (ADA) and modeled after existing methods, was utilized to clarify and codify the evidence that forms the basis for the recommendations. The level of evidence that supports each recommendation is listed after each recommendation using the letters A, B, C, or E.

Standards of Medical Care in Diabetes

Data Mining Practical Machine Learning Tools and Techniques

A study with low statistical power has a reduced chance of detecting a true effect, but it is less well appreciated that low power also reduces the likelihood that a statistically significant result reflects a true effect. Here, we show that the average statistical power of studies in the neurosciences is very low. The consequences of this include overestimates of effect size and low reproducibility of results. There are also ethical dimensions to this problem, as unreliable research is inefficient and wasteful. Improving reproducibility in neuroscience is a key priority and requires attention to well-established but often ignored methodological principles.

/pdf/power-failure-why-small-sample-size-undermines-the-4ftwwuexp7.pdf

Power failure: why small sample size undermines the reliability of neuroscience

Cluster Analysis: Everitt/Cluster Analysis

Context Current drug treatments for schizophrenia are inadequate for many patients, and despite 5 decades of drug discovery, all of the treatments rely on the same mechanism: dopamine D 2 receptor blockade. Understanding the pathophysiology of the disorder is thus likely to be critical to the rational development of new treatments for schizophrenia. Objective To investigate the nature of the dopaminergic dysfunction in schizophrenia using meta-analysis of in vivo studies. Data Sources The MEDLINE, EMBASE, and PsycINFO databases were searched for studies from January 1, 1960, to July 1, 2011. Study Selection A total of 44 studies were identified that compared 618 patients with schizophrenia with 606 controls, using positron emission tomography or single-photon emission computed tomography to measure in vivo striatal dopaminergic function. Data Extraction Demographic, clinical, and imaging variables were extracted from each study, and effect sizes were determined for the measures of dopaminergic function. Studies were grouped into those of presynaptic function and those of dopamine transporter and receptor availability. Sensitivity analyses were conducted to explore the consistency of effects and the effect of clinical and imaging variables. Data Synthesis There was a highly significant elevation (P  2/3 receptor availability (Cohen d = 0.26), but this was not evident in drug-naive patients and was influenced by the imaging approach used. Conclusions The locus of the largest dopaminergic abnormality in schizophrenia is presynaptic, which affects dopamine synthesis capacity, baseline synaptic dopamine levels, and dopamine release. Current drug treatments, which primarily act at D 2/3 receptors, fail to target these abnormalities. Future drug development should focus on the control of presynaptic dopamine synthesis and release capacity.

https://jamanetwork.com/journals/jamapsychiatry/articlepdf/1151090/yma120002_776_786.pdf

The nature of dopamine dysfunction in schizophrenia and what this means for treatment

Objectives Existing treatments for schizophrenia can improve positive symptoms, but it is unclear if they have any impact on negative symptoms. This meta-analysis was conducted to assess the efficacy of available treatments for negative symptoms in schizophrenia. Methods All randomized-controlled trials of interventions for negative symptoms in schizophrenia until December 2013 were retrieved; 168 unique and independent placebo-controlled trials were used. Negative symptom scores at baseline and follow-up, duration of illness, doses of medication, type of interventions, and sample demographics were extracted. Heterogeneity was addressed with the I (2) and Q statistic. Standardized mean difference in values of the Negative Symptom Rating Scale used in each study was calculated as the main outcome measure. Results 6503 patients in the treatment arm and 5815 patients in the placebo arm were included. No evidence of publication biases found. Most treatments reduced negative symptoms at follow-up relative to placebo: second-generation antipsychotics: -0.579 (-0.755 to -0.404); antidepressants: -0.349 (-0.551 to -0.146); combinations of pharmacological agents: -0.518 (-0.757 to -0.279); glutamatergic medications: -0.289 (-0.478 to -0.1); psychological interventions: -0.396 (-0.563 to -0.229). No significant effect was found for first-generation antipsychotics: -0.531 (-1.104 to 0.041) and brain stimulation: -0.228 (-0.775 to 0.319). Effects of most treatments were not clinically meaningful as measured on Clinical Global Impression Severity Scale. Conclusions and relevance Although some statistically significant effects on negative symptoms were evident, none reached the threshold for clinically significant improvement.

Treatments of Negative Symptoms in Schizophrenia: Meta-Analysis of 168 Randomized Placebo-Controlled Trials

Background. The aim was to critically appraise and synthesize the literature relating to set-shifting ability in eating disorders. PsycINFO, Medline, and Web of Science databases were searched to December 2005. Hand searching of eating-disorder journals and relevant reference sections was also undertaken. Method. The 15 selected studies contained both eating disorder and healthy control groups, and employed at least one of the following six neuropsychological measures of set-shifting ability ; Trail Making Test (TMT), Wisconsin Card Sort Test (WCST), Brixton task, Haptic Illusion, CatBat task, or the set-shifting subset of the Cambridge Neuropsychological Test Automated Battery (CANTAB). The outcome variable was performance on the set-shifting aspect of the task. Pooled standardized mean differences (effect sizes) were calculated. Results. TMT, WCST, CatBat and Haptic tasks had sufficient sample sizes for meta-analysis. These four tasks yielded acceptable pooled standardized effect sizes (0 . 36 ; TMT x1 . 05; Haptic) with moderate variation within studies (as measured by confidence intervals). The Brixton task showed a small pooled mean difference, and displayed more variation between sample results. The effect size for CANTAB set shifting was 0 . 17. Conclusion. Problems in set shifting as measured by a variety of neuropsychological tasks are present in people with eating disorders.

https://www.cambridge.org/core/services/aop-cambridge-core/content/view/S0033291707009877

A systematic review and meta-analysis of set-shifting ability in eating disorders.

UNLABELLED: Cannabis use is associated with an earlier age of onset of psychosis (AOP). However, the reasons for this remain debated. METHODS: We applied a Cox proportional hazards model to 410 first-episode psychosis patients to investigate the association between gender, patterns of cannabis use, and AOP. RESULTS: Patients with a history of cannabis use presented with their first episode of psychosis at a younger age (mean years = 28.2, SD = 8.0; median years = 27.1) than those who never used cannabis (mean years = 31.4, SD = 9.9; median years = 30.0; hazard ratio [HR] = 1.42; 95% CI: 1.16-1.74; P < .001). This association remained significant after controlling for gender (HR = 1.39; 95% CI: 1.11-1.68; P < .001). Those who had started cannabis at age 15 or younger had an earlier onset of psychosis (mean years = 27.0, SD = 6.2; median years = 26.9) than those who had started after 15 years (mean years = 29.1, SD = 8.5; median years = 27.8; HR = 1.40; 95% CI: 1.06-1.84; P = .050). Importantly, subjects who had been using high-potency cannabis (skunk-type) every day had the earliest onset (mean years = 25.2, SD = 6.3; median years = 24.6) compared to never users among all the groups tested (HR = 1.99; 95% CI: 1.50- 2.65; P < .0001); these daily users of high-potency cannabis had an onset an average of 6 years earlier than that of non-cannabis users. CONCLUSIONS: Daily use, especially of high-potency cannabis, drives the earlier onset of psychosis in cannabis users.

/pdf/daily-use-especially-of-high-potency-cannabis-drives-the-3icul0l9yd.pdf

Daniel Stahl

Papers

Cluster Analysis: Everitt/Cluster Analysis

The nature of dopamine dysfunction in schizophrenia and what this means for treatment

Treatments of Negative Symptoms in Schizophrenia: Meta-Analysis of 168 Randomized Placebo-Controlled Trials

A systematic review and meta-analysis of set-shifting ability in eating disorders.

Daily Use, Especially of High-Potency Cannabis, Drives the Earlier Onset of Psychosis in Cannabis Users.