Accurate Patient-Specific Machine Learning Models of Glioblastoma Invasion Using Transfer Learning.

BACKGROUND AND PURPOSE: MR imaging-based modeling of tumor cell density can substantially improve targeted treatment of glioblastoma. Unfortunately, interpatient variability limits the predictive ability of many modeling approaches. We present a transfer learning method that generates individualized patient models, grounded in the wealth of population data, while also detecting and adjusting for interpatient variabilities based on each patient's own histologic data. MATERIALS AND METHODS: We recruited patients with primary glioblastoma undergoing image-guided biopsies and preoperative imaging, including contrast-enhanced MR imaging, dynamic susceptibility contrast MR imaging, and diffusion tensor imaging. We calculated relative cerebral blood volume from DSC-MR imaging and mean diffusivity and fractional anisotropy from DTI. Following image coregistration, we assessed tumor cell density for each biopsy and identified corresponding localized MR imaging measurements. We then explored a range of univariate and multivariate predictive models of tumor cell density based on MR imaging measurements in a generalized one-model-fits-all approach. We then implemented both univariate and multivariate individualized transfer learning predictive models, which harness the available population-level data but allow individual variability in their predictions. Finally, we compared Pearson correlation coefficients and mean absolute error between the individualized transfer learning and generalized one-model-fits-all models. RESULTS: Tumor cell density significantly correlated with relative CBV (r = 0.33, P < .001), and T1-weighted postcontrast (r = 0.36, P < .001) on univariate analysis after correcting for multiple comparisons. With single-variable modeling (using relative CBV), transfer learning increased predictive performance (r = 0.53, mean absolute error = 15.19%) compared with one-model-fits-all (r = 0.27, mean absolute error = 17.79%). With multivariate modeling, transfer learning further improved performance (r = 0.88, mean absolute error = 5.66%) compared with one-model-fits-all (r = 0.39, mean absolute error = 16.55%). CONCLUSIONS: Transfer learning significantly improves predictive modeling performance for quantifying tumor cell density in glioblastoma.

The new Sheffield risk and benefit tables for the elderly.

Several charts or tables are used to guide treatment in primary prevention of cardiovascular disease (CVD). These usually relate to patients up to 75 years of age, leaving older patients without guidance. Most also present this information as risk, leaving patients to estimate the benefit of treatment and decide whether it is worthwhile. We present tables to display both CVD risk and benefit from treatment in the elderly. A systematic review identified CVD risk functions for the elderly. The Dubbo study of older patients' 5-year CVD risk equation was deemed most appropriate, due to the population studied, endpoints observed and risk factors recorded. By dichotomizing most risk factors, we produced a new risk table in the form of the original 'Sheffield table'. Risk is calculated by selecting the appropriate table for gender and the appropriate cell from the rows and columns, representing age and risk factor contributors, respectively. Total cholesterol above a cell value corresponds to a 20 or 40% 10-year CVD risk. A simple risk scoring system was then derived from the Dubbo equation. Calculation of risk score requires knowledge of a patient's simple demographics, systolic blood pressure and total and high-density lipoprotein cholesterol. Positive integers corresponding to level of risk for each contributing factor are then added together to give a final risk score. A Markov chain model was produced based on the Dubbo derived risk and relative risk reductions from published meta-analyses of 3-hydroxy-3-methyl-glutaryl-CoA reductase inhibitors (statins) and anti-hypertensive treatment. Using this model, individual scores were mapped to likely benefit from treatment in terms of disease free years. Our risk table provides a simple means for calculating risk in the elderly, to two major thresholds, while the benefit table explores the concept of presenting benefit of taking CVD-preventing medication.

A novel size-selective airborne particle size fractionating instrument for health risk evaluation.

Health risks associated with the inhalation of airborne particles are known to be influenced by particle size. Studies have shown that certain nanoparticles, with diameters <100 nm, have increased toxicity relative to larger particles of the same substance. A reliable, size-resolving sampler able to collect a wide range of particle sizes, including particles with sizes in the nanometre range, would be beneficial in investigating health risks associated with the inhalation of airborne particles. A review of current aerosol samplers used for size-resolved collection of airborne particles highlighted a number of limitations. These could be overcome by combining an inertial deposition impactor with a diffusion collector in a single device. Verified theories of diffusion and inertial deposition suggested an optimal design and operational regime. The instrument was designed for analysing mass distribution functions. Calibration was carried out using a number of recognized techniques. The sampler was tested in the field by collecting size-resolved samples of lead containing aerosols present at workplaces in factories producing crystal glass. The mass deposited on each screen proved sufficient to be detected and measured by an appropriate analytical technique. Mass concentration distribution functions of lead were produced. The nanofraction of lead in air varied from 10 to 70% by weight of total lead.

No evidence that panic attacks are associated with the white coat effect in hypertension.

The authors aimed to determine whether hypertensive patients with panic attacks or panic disorder have a larger white coat effect (difference between clinic blood pressure measured under standard conditions and mean daytime ambulatory blood pressure) than hypertensive patients without panic attacks. White coat effect was compared in a hospital hypertension clinic between 24 patients with panic attacks in the previous 6 months (12 with panic disorder) and 23 hypertensive controls. There were no significant differences between cases and controls in clinic blood pressure, mean daytime ambulatory blood pressure, or white coat effect (18/3 vs. 19/6 mm Hg; difference for systolic, -1.9 mm Hg; 95% confidence interval, -15.8 to +12.0; difference for diastolic, -3.0 mm Hg; 95% confidence interval, -10.2 to +4.3). Comparing only patients with panic disorder with controls, there were again no significant differences in clinic blood pressure, mean daytime ambulatory blood pressure, or white coat effect. This study provides no evidence for an exaggerated white coat effect in hypertensive patients who have experienced panic attacks or panic disorder. However, only larger studies could exclude differences in white coat effect <12/4 mm Hg, or an exaggerated white coat effect in a minority of patients with panic attacks.

Statins for primary prevention: at what coronary risk is safety assured?

AIMS: Increasingly HMG CoA reductase inhibitors (statins) are being used for primary prevention of vascular disease in patients with a raised cholesterol but at low absolute risk of coronary heart disease (CHD). This study uses clinical trial results to explore the limits of absolute safety for statin use in such patients. METHODS: The major placebo controlled statin outcome trials were identified by automated and manual literature searches. Principal results including all cause mortality in placebo and intervention groups and baseline values of standard coronary risk factors were abstracted for each trial. For the trials identified the reduction in overall mortality with statin treatment for each study was regressed against the underlying CHD risk of the population recruited into that trial using a statistically robust method. RESULTS: The regression line describing the relationship between mortality benefit and risk suggests that statin use could be associated with an increase in mortality of 1% in 10 years. This would be sufficiently large to negate statin's beneficial effect on CHD mortality in patients with a CHD event risk less than 13% over 10 years. CONCLUSIONS: Absolute safety of statins has not been demonstrated for patients at low risk of CHD. Patients absolute risk of CHD should be calculated before starting statin treatment for primary prevention. Extensions of such treatment to low risk patients should await further evidence of safety.

Realizing the potential of empowerment: the impact of a feedback intervention on the performance of complex technology.

An empowerment initiative involving enhanced fault-management responsibility for operators of complex technology had not led to expected increases in performance, and investigations suggested that this was due to a lack of appropriate feedback. Thus, a feedback intervention was designed to provide specific, timely feedback on operator-correctable faults. It was hypothesized that the intervention would increase operator self-reliance in operating complex technology and promote system performance. Moreover, given the feedback was continuous from the point of intervention, it was predicted that gains would increase over time. Time series analysis of data on engineer call-outs (self-reliance) and machine utilization (performance) showed clear positive effects of the feedback intervention, with call-outs also showing progressive improvement. Self-report data showed no change over time in motivation, but an increase in knowledge dissemination and a reduction in the likelihood of making expensive mistakes. There were no detrimental effects on operator well being. Implications for theory and practice in the management of complex technology are discussed.

Population implications of lipid lowering for prevention of coronary heart disease: data from the 1995 Scottish health survey.

OBJECTIVE: To determine the proportion of the population, firstly, with cholesterol >/= 5.0 mmol/l and, secondly, with any cholesterol concentration, who might benefit from statin treatment for the following: secondary prevention of coronary heart disease (CHD); primary prevention at CHD risk 30%, 20%, 15%, and 6% over 10 years; and primary prevention at projected CHD risk 20% over 10 years (CHD risk at age 60 years if actual age < 60 years). SUBJECTS: Random stratified sample of 3963 subjects aged 35-64 years from the Scottish health survey 1995. RESULTS: For secondary prevention 7.8% (95% confidence interval (CI) 6.9% to 8.6%) of the population with cholesterol >/= 5.0 mmol/l would benefit from statins. For primary prevention, the prevalence of people at CHD risk 30%, 20%, 15%, and 6% over 10 years is 1.5% (95% CI 1.2% to 1.9%), 5.4% (95% CI 4.7% to 6.1%), 9.7% (95% CI 8.8% to 10.6%), and 32.9% (95% CI 31.5% to 34.4%), respectively. At projected CHD risk 20% over 10 years, 12.4% (95% CI 11.4% to 13.5%) would be treated with statins. Removing the 5.0 mmol/l cholesterol threshold makes little difference to population prevalence at high CHD risk. CONCLUSIONS: Statin treatment would be required for 7.8% of the population for secondary prevention. For primary prevention, among other factors, guidelines should take into account the number of patients needing treatment at different levels of CHD risk when choosing the CHD risk to target. The analysis supports a policy of targeting treatment at CHD risk 30% over 10 years as a minimum, as recommended in current British guidelines, with a move to treating at CHD risk 15% over 10 years as resources permit.

Is coronary risk an accurate surrogate for cardiovascular risk for treatment decisions in mild hypertension? A population validation.

OBJECTIVE: To examine the relationship between coronary (CHD) and cardiovascular (CVD) risk in patients with uncomplicated mild hypertension and to determine the accuracy of using CHD risk > or = 15% over 10 years to identify for antihypertensive treatment those patients with CVD risk > or = 20% over 10 years as advised in recent British guidelines. DESIGN: Comparison of decisions made using CHD risk > or = 15% over 10 years calculated by the Framingham risk function and estimated using a simple table with CVD risk > or = 20% over 10 years. SETTING: British population. SUBJECTS: People aged 35-64 years with uncomplicated mild systolic hypertension (SBP 140-159 mmHg, n = 624) from the 1995 Scottish Health Survey. MAIN OUTCOME MEASURES: Relationship between CHD and CVD risk. Sensitivity, specificity, positive and negative predictive values (PPV and NPV). RESULTS: CHD risk 15% over 10 years was equivalent to CVD risk 21% over 10 years. Exact CHD risk > or = 15% over 10 years had sensitivity 79%, specificity 98%, PPV 94% and NPV 93% in detecting CVD risk > or = 20% over 10 years. Use of the table to estimate CHD risk > or = 15% over 10 years gave sensitivity 88%, specificity 90%, PPV 76% and NPV 95%. CONCLUSION: CHD risk appears acceptably accurate for targeting treatment in mild hypertension. The risk assessment table, which slightly overestimates CHD risk, was more sensitive in identifying patients with CVD risk > or = 20% over 10 years and may be preferable to using exact CHD risk. European guidelines which suggest targeting treatment for mild hypertension at CHD risk > or = 20% over 10 years are over-conservative compared with British guidelines.

Aspirin for primary prevention of coronary heart disease: safety and absolute benefit related to coronary risk derived from meta-analysis of randomised trials.

OBJECTIVE: To determine the cardiovascular and coronary risk thresholds at which aspirin for primary prevention of coronary heart disease is safe and worthwhile. DESIGN: Meta-analysis of four randomised controlled trials of aspirin for primary prevention. The benefit and harm from aspirin treatment were examined to determine: (1) the cardiovascular and coronary risk threshold at which benefit in prevention of myocardial infarction exceeds harm from significant bleeding; and (2) the absolute benefit expressed as number needed to treat (NNT) for aspirin net of cerebral haemorrhage and other bleeding complications at different levels of coronary risk. MAIN OUTCOME MEASURES: Benefit from aspirin, expressed as reduction in cardiovascular events, myocardial infarctions, strokes, and total mortality; harm caused by aspirin in relation to significant bleeds and major haemorrhages. RESULTS: Aspirin for primary prevention significantly reduced all cardiovascular events by 15% (95% confidence interval (CI) 6% to 22%) and myocardial infarctions by 30% (95% CI 21% to 38%), and non-significantly reduced all deaths by 6% (95% CI -4% to 15%). Aspirin non-significantly increased strokes by 6% (95% CI -24% to 9%) and significantly increased bleeding complications by 69% (95% CI 38% to 107%). The risk of major bleeding balanced the reduction in cardiovascular events when cardiovascular event risk was 0.22%/year. The upper 95% CI for this estimate suggests that harm from aspirin is unlikely to outweigh benefit provided the cardiovascular event risk is 0.8%/year, equivalent to a coronary risk of 0.6%/year. At coronary event risk 1.5%/year, the five year NNT was 44 to prevent a myocardial infarction, and 77 to prevent a myocardial infarction net of any important bleeding complication. At coronary event risk 1%/year the NNT was 67 to prevent a myocardial infarction, and 182 to prevent a myocardial infarction net of important bleeding. CONCLUSIONS: Aspirin treatment for primary prevention is safe and worthwhile at coronary event risk >/= 1.5%/year; safe but of limited value at coronary risk 1%/year; and unsafe at coronary event risk 0.5%/year. Advice on aspirin for primary prevention requires formal accurate estimation of absolute coronary event risk.

Debate: at what level of coronary heart disease risk should a statin be prescribed?

3-Hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins) are effective treatments for the primary and secondary prevention of coronary heart disease, but an outstanding issue is determining who should have such treatment. The benefit from treatment with statins appears to be proportional to the underlying risk of coronary heart disease and independent of the factors increasing risk. Most benefit will therefore be achieved by treating people at increased risk of coronary heart disease. Statins reduce coronary morbidity even when the risk of coronary heart disease is relatively low (6% over 10 years), but reduction in all-cause mortality, the true measure of safety has been shown only when the risk of a major coronary heart disease event is 15% over 10 years or greater. At this level of risk patients appear willing to take treatment to gain the benefit expected from statin treatment, and the cost effectiveness of statin treatment is within the range accepted for other treatments. The major impediments to the systematic introduction of statin treatment at this level of risk are the very high overall cost and the large workload in countries like Britain, where the population risk of coronary heart disease is high. For this reason, recent British guidelines correctly advise statin treatment for secondary prevention and primary prevention when the 10 year coronary heart disease risk is 30% or greater as the first priority, moving to a lower coronary heart disease threshold for primary prevention only when resources permit.

Lean production teams and health in garment manufacture.

The impact of lean production on psychological health was assessed by comparing lean production teams in garment manufacture with a traditional system for making similar garments. Work design characteristics were examined as mediators of the impact of work organization on health. Findings indicate both positive and negative direct effects of teamworking on aspects of autonomy, work demands, and social climate. In turn, both positive and negative direct effects of work design on psychological health were found, which combine to give no overall difference between the groups. This study suggests that the balance between positive and negative effects of lean production teamworking depends on management choices in the form of work design.

Coronary and cardiovascular risk estimation for primary prevention: validation of a new Sheffield table in the 1995 Scottish health survey population.

OBJECTIVE: To examine the accuracy of a new version of the Sheffield table designed to aid decisions on lipids screening and detect thresholds for risk of coronary heart disease needed to implement current guidelines for primary prevention of cardiovascular disease. DESIGN: Comparison of decisions made on the basis of the table with absolute risk of coronary heart disease or cardiovascular disease calculated by the Framingham risk function. The decisions related to statin treatment when coronary risk is >/=30% over 10 years; aspirin treatment when the risk is >/=15% over 10 years; and the treatment of mild hypertension when the cardiovascular risk is >/=20% over 10 years. SETTING: The table is designed for use in general practice. SUBJECTS: Random sample of 1000 people aged 35-64 years from the 1995 Scottish health survey. MAIN OUTCOME MEASURES: Sensitivity, specificity, and positive and negative predictive values of the table. RESULTS: 13% of people had a coronary risk of >/=15%, and 2. 2% a risk of >/=30%, over 10 years. 22% had mild hypertension (systolic blood pressure 140-159 mm Hg). The table indicated lipids screening for everyone with a coronary risk of >/=15% over 10 years, for 95% of people with a ratio of total cholesterol to high density lipoprotein cholesterol of >/=8.0, but for <50% with a coronary risk of <5% over 10 years. Sensitivity and specificity were 97% and 95% respectively for a coronary risk of >/=15% over 10 years; 82% and 99% for a coronary risk of >/=30% over 10 years; and 88% and 90% for a cardiovascular risk of >/=20% over 10 years in mild hypertension. CONCLUSION: The table identifies all high risk people for lipids screening, reduces screening of low risk people by more than half, and ensures that treatments are prescribed appropriately to those at high risk, while avoiding inappropriate treatment of people at low risk.