A machine learning approach using EEG data to predict response to SSRI treatment for major depressive disorder

Objective

The problem of identifying, in advance, the most effective treatment agent for various psychiatric conditions remains an elusive goal. To address this challenge, we investigate the performance of the proposed machine learning (ML) methodology (based on the pre-treatment electroencephalogram (EEG)) for prediction of response to treatment with a selective serotonin reuptake inhibitor (SSRI) medication in subjects suffering from major depressive disorder (MDD).

Methods

A relatively small number of most discriminating features are selected from a large group of candidate features extracted from the subject’s pre-treatment EEG, using a machine learning procedure for feature selection. The selected features are fed into a classifier, which was realized as a mixture of factor analysis (MFA) model, whose output is the predicted response in the form of a likelihood value. This likelihood indicates the extent to which the subject belongs to the responder vs. non-responder classes. The overall method was evaluated using a “leave-n-out” randomized permutation cross-validation procedure.

Results

A list of discriminating EEG biomarkers (features) was found. The specificity of the proposed method is 80.9% while sensitivity is 94.9%, for an overall prediction accuracy of 87.9%. There is a 98.76% confidence that the estimated prediction rate is within the interval [75%, 100%].

Conclusions

These results indicate that the proposed ML method holds considerable promise in predicting the efficacy of SSRI antidepressant therapy for MDD, based on a simple and cost-effective pre-treatment EEG.

Significance

The proposed approach offers the potential to improve the treatment of major depression and to reduce health care costs.     

Biomarkers predicting sepsis in polytrauma patients: Current evidence

Introduction

Major trauma still represents one of the leading causes of death in the first four decades of life. Septic complications represent the predominant causes of late death (45% of overall mortality) in polytrauma patients. The ability of clinicians to early differentiate between systemic inflammatory response syndrome (SIRS) and sepsis is demonstrated to improve clinical outcome and mortality. The identification of an “ideal” biomarker able to early recognize incoming septic complications in trauma patients is still a challenge for researchers.

Aim

To evaluate the existing evidence regarding the role of biomarkers to predict or facilitate early diagnosis of sepsis in trauma patients, trying to compile some recommendations for the clinical setting.

Methods

An Internet-based search of the MEDLINE, EMBASE and Cochrane Library databases was performed using the search terms: “Biomarkers”, “Sepsis” and “Trauma” in various combinations. The methodological quality of the included studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies Checklist (QUADAS). After data extraction, the level of evidence available for each bio-marker was rated and presented using the “best-evidence synthesis” method, in line with the US Agency for Healthcare Research and Quality.

Results

Thirty studies were eligible for the final analysis: 13 case–control studies and 17 cohort studies. The “strong evidence” available demonstrated the potential use of procalcitonin as an early indicator of post-traumatic septic complications and reported the inability of c-reactive protein (CRP) to specifically identify infective complications. Moderate, conflicting and limited evidence are available for the other 31 biomarkers.

Conclusion

Several biomarkers have been evaluated for predicting or making early diagnosis of sepsis in trauma patients. Current evidence does not support the use of a single biomarker in diagnosing sepsis. However, procalcitonin trend was found to be useful in early identification of post-traumatic septic course and its use is suggested (Recommendation Grade: B) in clinical practice.     

Monitoring Biomarkers After Pediatric Heart Surgery: A New Paradigm on the Horizon

Until recently, risk scoring systems for adult patients consisted of only clinical criteria. Currently, we are experiencing an abundant surge of literature integrating a wide range of biomarker arrays to clinical criteria in assessing the risk in an adult. In fact, novel risk scoring systems such as Reynolds criteria have emerged by combining the validated biomarkers to the traditional risk factors. Novel biomarkers potentially improve clinical management of cardiovascular disease, but there are gaps in understanding their role during childhood. The reason might be related to relatively lower prevalence of cardiovascular disease in children compared to the adult population. One exceptional group is the children with congenital heart disease. Recent studies indicate that novel biomarkers can alert the clinician in a timely manner about neurological and myocardial injury and their inflammatory consequences. Current technologies enable us to measure several biomarkers using only a few microliters of plasma. The preliminary studies show that novel biomarkers in addition to the traditionally studied biomarkers can help the clinician to identify children at high risk following pediatric heart surgery. Future studies are needed to confirm the role of biomarkers in monitoring children after cardiopulmonary bypass.     

Novel biomarkers for risk stratification and identification of lifethreatening cardiovascular disease: troponin and beyond

Chest pain and other symptoms that may represent acute coronary syndromes (ACS) are common reasons for emergency department (ED) presentations, accounting for over six million visits annually in the United States [1]. Chest pain is the second most common ED presentation in the United States. Delays in diagnosis and inaccurate risk stratification of chest pain can result in serious morbidity and mortality from ACS, pulmonary embolism (PE), aortic dissection and other serious pathology. Because of the high morbidity, mortality, and liability issues associated with both recognized and unrecognized cardiovascular pathology, an aggressive approach to the evaluation of this patient group has become the standard of care. Clinical history, physical examination and electrocardiography have a limited diagnostic and prognostic role in the evaluation of possible ACS, PE, and aortic dissection, so clinicians continue to seek more accurate means of risk stratification. Recent advances in diagnostic imaging techniques particularly computed-tomography of the coronary arteries and aorta, have significantly improved our ability to diagnose life-threatening cardiovascular disease. In an era where health care utilization and cost are major considerations in how disease is managed, it is crucial to riskstratify patients quickly and efficiently. Historically, biomarkers have played a significant role in the diagnosis and risk stratification of several cardiovascular disease states including myocardial infarction, congestive heart failure, and pulmonary embolus. Multiple biomarkers have shown early promise in answering questions of risk stratification and early diagnosis of cardiovascular pathology however many do not yet have wide clinical availability. The goal of this review will be to discuss these novel biomarkers and describe their potential role in direct patient care.     

Clinical indications for analysis of Alzheimer's disease CSF biomarkers

The cerebrospinal fluid (CSF) biomarkers β-amyloid_1-42 (Aβ_1-42), total tau protein (T-tau) and hyperphosphorylated tau (P-tau_181P) are well-validated and are increasingly used in clinical practice as an affirmative diagnostic tool for Alzheimer's disease (AD). These biomarkers have also been implemented in the revised diagnostic criteria of AD. The combination of the CSF biomarkers Aβ_1-42, T-tau and P-tau_181P results in high levels of sensitivity, specificity and diagnostic accuracy for discriminating AD from controls (including psychiatric disorders like depression). These biomarkers can be applied for diagnosing AD in the prodromal phase of the disease (mild cognitive impairment). In case of doubt between vascular dementia (VaD) or mixed AD-VaD pathology in dementia patients, the determination of CSF Aβ_1-42, T-tau and P-tau_181P levels is of help to confirm or exclude the AD component in the pathophysiology of the dementia syndrome. However, their discriminatory power for the differential diagnosis of dementia is suboptimal. Other CSF biomarkers like Aβ_1-40, and those that are reflective of the pathology of non-AD dementias, could improve the accuracy of differential dementia diagnosis. The added differential diagnostic value of the CSF biomarkers Aβ_1-42, T-tau and P-tau_181P could lie within those cases in which the routine clinical diagnostic work-up is not able to discriminate between AD or non-AD dementias. In summary, the CSF biomarkers Aβ_1-42, T-tau and P-tau_181P can be used in clinical practice to discriminate AD from healthy aging (including psychiatric disorders like depression), to diagnose AD in its prodromal phase or in atypical forms with prominent non-memory impairment, to identify AD in patients with mixed pathologies and in case of an ambiguous (AD versus non-AD) dementia diagnosis.     

Metabolism and metabolomics of opiates: A long way of forensic implications to unravel

Opium poppy has important medical, socioeconomic, forensic and political implications. More than 80 benzylisoquinoline alkaloids have been described, many of them with relevant therapeutic properties such as morphine, codeine, papaverine and noscapine. Heroin, a semi-synthetic drug produced from morphine is a worldwide serious cause of morbidity and mortality. Heroin dependence is complex phenomenon with environmental and genetic influence, and several biomarkers of exposure have been proposed. This work aims to review the metabolism and metabolomics of opiates with particular interest on their relevance as potential clinical and forensic antemortem and postmortem biomarkers. It is known that the heroin is mainly a prodrug that is rapidly deacetylated in blood to its active metabolite, 6-acetylmorphine, which is then subsequently slowly deacetylated to morphine. Therefore, 6-acetylmorphine has been used as the main target metabolite to prove heroin abuse in clinical, but mostly in forensic routine. Nevertheless, its applicability is limited due to the reduced detection window. Therefore, morphine (and its metabolites morphine-3-glucuronide and morphine-6-glucuronide), codeine, codeine-6-glucuronide, 6-acetylcodeine, noscapine (and its metabolites meconine, desmethylmeconine, and cotarnine), papaverine (and its metabolites 6-desmethylpapaverine, hydroxypapaverine, dihydroxypapaverine, 6-desmethylpapaverine-glucuronide) and thebaine (and acetylthebaol and the non-acetylated analog thebaol) have been additionally recommended to obtain the most reliable results possible. More recently, the identification by metabolomics analysis of several endogenous compounds offered an alternative approach of significant importance to uncover toxic effects. Profound alterations in the neurotransmitters levels and energy and amino acid metabolism have been reported with l-tryptophan, 5-hydroxytryptamine and 5-hydroxyindoleacetate being suggested as potential non-specific biomarkers of long-term heroin addiction. These endogenous metabolic profiles and exogenous components that together comprise the exposome will certainly help to uncover metabolic disturbances and patterns that may be associate to addiction with relevant clinical and forensic implications.     

MYC-induced nuclear antigen (MINA) and preeclampsia

Background: Inadequate trophoblast invasion and the subsequent inflammatory response have been implicated in preeclampsia (PE) pathogenesis. Because MYC-induced nuclear antigen (MINA) gene expression is involved in cell proliferation and differentiation, inflammatory response modulation, and the unpaired regulation of which is associated with human diseases, we sought to investigate the connection between MINA and PE.

Objective: The aim of this study was to evaluate the possible relationship between the MINA rs4857304 variant and susceptibility to PE development as well as to estimate placental MINA gene expression and its association with PE.

Methods: About 242 pregnant women (126 PE cases and 116 controls) were included. MINA genotyping and gene expression were evaluated by quantitative real-time polymerase chain reaction using TaqMan probes.

Results: The G/G genotype of the MINA rs4857304 variant was associated with severe PE (p = 0.027, OR = 1.8, 95% CI = 1.8–3.2). Carriers of one G allele of the MINA rs4857304 variant exhibited a 1.7-fold increased risk of severe PE (p = 0.029, 95% CI = 1.1–3.0). MINA was underexpressed in preeclamptic placentas and MINA expression differed between the mild and severe PE groups. Differences in the expression levels of MINA were found among women with the T/T genotype of the rs4857304 polymorphism and carriers of at least one G allele (p = 0.024).

Conclusion: PE and its severity are associated with the underexpression of placental MINA, and the G/G genotype of the MINA rs4857304 variant may modify the risk of severe PE among the PE cases evaluated.