Previous research in clinical electroencephalography (EEG) has demonstrated that reduction of alpha frequency (8–13 Hz) EEG activity may have particular relevance to the negative symptoms of schizophrenia. Repetitive Transcranial Magnetic Stimulation (rTMS) was utilized to investigate this relationship by assessing the therapeutic effects of stimulation set individually at each subject's peak alpha frequency (αTMS). Twenty-seven subjects, with predominantly negative symptom schizophrenia, received 2 weeks of daily treatment with either αTMS, 3 Hz, 20 Hz, or sham stimulation bilaterally over the dorsolateral prefrontal cortex. Individualized αTMS demonstrated a significantly larger (F _3,33 = 4.7, p = .007) therapeutic effect (29.6% reduction in negative symptoms) than the other 3 conditions (< 9%). Furthermore, these clinical improvements were found to be highly correlated (r = 0.86, p = .001) with increases (34%) in frontal alpha amplitude following αTMS. These results affirm that the resonant features of alpha frequency EEG play an important role in the pathophysiology of schizophrenia and merit further investigation as a particularly efficacious frequency for rTMS treatments.

On September 18, 2007, a collaborative session between the International Society for CNS Clinical Trials and Methodology and the International Society for CNS Drug Development was held in Brussels, Belgium. Both groups, with membership from industry, academia, and governmental and nongovernmental agencies, have been formed to address scientific, clinical, regulatory, and methodological challenges in the development of central nervous system therapeutic agents. The focus of this joint session was the apparent diminution of drug-placebo differences in recent multicenter trials of antipsychotic medications for schizophrenia. To characterize the nature of the problem, some presenters reported data from several recent trials that indicated higher rates of placebo response and lower rates of drug response (even to previously established, comparator drugs), when compared with earlier trials. As a means to identify the possible causes of the problem, discussions covered a range of methodological factors such as participant characteristics, trial designs, site characteristics, clinical setting (inpatient vs outpatient), inclusion/exclusion criteria, and diagnostic specificity. Finally, possible solutions were discussed, such as improving precision of participant selection criteria, improving assessment instruments and/or assessment methodology to increase reliability of outcome measures, innovative methods to encourage greater subject adherence and investigator involvement, improved rater training and accountability metrics at clinical sites to increase quality assurance, and advanced methods of pharmacokinetic/pharmacodynamic modeling to optimize dosing prior to initiating large phase 3 trials. The session closed with a roundtable discussion and recommendations for data sharing to further explore potential causes and viable solutions to be applied in future trials.

Machine learning approaches have been used for the automatic detection of Parkinson’s disease with voice recordings being the most used data type due to the simple and non-invasive nature of acquiring such data. Although voice recordings captured via telephone or mobile devices allow much easier and wider access for data collection, current conflicting performance results limit their clinical applicability. This study has two novel contributions. First, we show the reliability of personal telephone-collected voice recordings of the sustained vowel /a/ in natural settings by collecting samples from 50 people with specialist-diagnosed Parkinson’s disease and 50 healthy controls and applying machine learning classification with voice features related to phonation. Second, we utilize a novel application of a pre-trained convolutional neural network (Inception V3) with transfer learning to analyze the spectrograms of the sustained vowel from these samples. This approach considers speech intensity estimates across time and frequency scales rather than collapsing measurements across time. We show the superiority of our deep learning model for the task of classifying people with Parkinson’s disease as distinct from healthy controls.