Mortality Risk Assessment Using Deep Learning-Based Frequency Analysis of EEG and EOG in Sleep.

Kristjánsson TÓ ，Stone KL ，Sorensen HBD ，Brink-Kjaer A ，Mignot E ，Jennum P ... -  《-》

Comparison of Two Modern Survival Prediction Tools, SORG-MLA and METSSS, in Patients With Symptomatic Long-bone Metastases Who Underwent Local Treatment With Surgery Followed by Radiotherapy and With Radiotherapy Alone.

Survival estimation for patients with symptomatic skeletal metastases ideally should be made before a type of local treatment has already been determined. Currently available survival prediction tools, however, were generated using data from patients treated either operatively or with local radiation alone, raising concerns about whether they would generalize well to all patients presenting for assessment. The Skeletal Oncology Research Group machine-learning algorithm (SORG-MLA), trained with institution-based data of surgically treated patients, and the Metastases location, Elderly, Tumor primary, Sex, Sickness/comorbidity, and Site of radiotherapy model (METSSS), trained with registry-based data of patients treated with radiotherapy alone, are two of the most recently developed survival prediction models, but they have not been tested on patients whose local treatment strategy is not yet decided. (1) Which of these two survival prediction models performed better in a mixed cohort made up both of patients who received local treatment with surgery followed by radiotherapy and who had radiation alone for symptomatic bone metastases? (2) Which model performed better among patients whose local treatment consisted of only palliative radiotherapy? (3) Are laboratory values used by SORG-MLA, which are not included in METSSS, independently associated with survival after controlling for predictions made by METSSS? Between 2010 and 2018, we provided local treatment for 2113 adult patients with skeletal metastases in the extremities at an urban tertiary referral academic medical center using one of two strategies: (1) surgery followed by postoperative radiotherapy or (2) palliative radiotherapy alone. Every patient's survivorship status was ascertained either by their medical records or the national death registry from the Taiwanese National Health Insurance Administration. After applying a priori designated exclusion criteria, 91% (1920) were analyzed here. Among them, 48% (920) of the patients were female, and the median (IQR) age was 62 years (53 to 70 years). Lung was the most common primary tumor site (41% [782]), and 59% (1128) of patients had other skeletal metastases in addition to the treated lesion(s). In general, the indications for surgery were the presence of a complete pathologic fracture or an impending pathologic fracture, defined as having a Mirels score of ≥ 9, in patients with an American Society of Anesthesiologists (ASA) classification of less than or equal to IV and who were considered fit for surgery. The indications for radiotherapy were relief of pain, local tumor control, prevention of skeletal-related events, and any combination of the above. In all, 84% (1610) of the patients received palliative radiotherapy alone as local treatment for the target lesion(s), and 16% (310) underwent surgery followed by postoperative radiotherapy. Neither METSSS nor SORG-MLA was used at the point of care to aid clinical decision-making during the treatment period. Survival was retrospectively estimated by these two models to test their potential for providing survival probabilities. We first compared SORG to METSSS in the entire population. Then, we repeated the comparison in patients who received local treatment with palliative radiation alone. We assessed model performance by area under the receiver operating characteristic curve (AUROC), calibration analysis, Brier score, and decision curve analysis (DCA). The AUROC measures discrimination, which is the ability to distinguish patients with the event of interest (such as death at a particular time point) from those without. AUROC typically ranges from 0.5 to 1.0, with 0.5 indicating random guessing and 1.0 a perfect prediction, and in general, an AUROC of ≥ 0.7 indicates adequate discrimination for clinical use. Calibration refers to the agreement between the predicted outcomes (in this case, survival probabilities) and the actual outcomes, with a perfect calibration curve having an intercept of 0 and a slope of 1. A positive intercept indicates that the actual survival is generally underestimated by the prediction model, and a negative intercept suggests the opposite (overestimation). When comparing models, an intercept closer to 0 typically indicates better calibration. Calibration can also be summarized as log(O:E), the logarithm scale of the ratio of observed (O) to expected (E) survivors. A log(O:E) > 0 signals an underestimation (the observed survival is greater than the predicted survival); and a log(O:E) < 0 indicates the opposite (the observed survival is lower than the predicted survival). A model with a log(O:E) closer to 0 is generally considered better calibrated. The Brier score is the mean squared difference between the model predictions and the observed outcomes, and it ranges from 0 (best prediction) to 1 (worst prediction). The Brier score captures both discrimination and calibration, and it is considered a measure of overall model performance. In Brier score analysis, the "null model" assigns a predicted probability equal to the prevalence of the outcome and represents a model that adds no new information. A prediction model should achieve a Brier score at least lower than the null-model Brier score to be considered as useful. The DCA was developed as a method to determine whether using a model to inform treatment decisions would do more good than harm. It plots the net benefit of making decisions based on the model's predictions across all possible risk thresholds (or cost-to-benefit ratios) in relation to the two default strategies of treating all or no patients. The care provider can decide on an acceptable risk threshold for the proposed treatment in an individual and assess the corresponding net benefit to determine whether consulting with the model is superior to adopting the default strategies. Finally, we examined whether laboratory data, which were not included in the METSSS model, would have been independently associated with survival after controlling for the METSSS model's predictions by using the multivariable logistic and Cox proportional hazards regression analyses. Between the two models, only SORG-MLA achieved adequate discrimination (an AUROC of > 0.7) in the entire cohort (of patients treated operatively or with radiation alone) and in the subgroup of patients treated with palliative radiotherapy alone. SORG-MLA outperformed METSSS by a wide margin on discrimination, calibration, and Brier score analyses in not only the entire cohort but also the subgroup of patients whose local treatment consisted of radiotherapy alone. In both the entire cohort and the subgroup, DCA demonstrated that SORG-MLA provided more net benefit compared with the two default strategies (of treating all or no patients) and compared with METSSS when risk thresholds ranged from 0.2 to 0.9 at both 90 days and 1 year, indicating that using SORG-MLA as a decision-making aid was beneficial when a patient's individualized risk threshold for opting for treatment was 0.2 to 0.9. Higher albumin, lower alkaline phosphatase, lower calcium, higher hemoglobin, lower international normalized ratio, higher lymphocytes, lower neutrophils, lower neutrophil-to-lymphocyte ratio, lower platelet-to-lymphocyte ratio, higher sodium, and lower white blood cells were independently associated with better 1-year and overall survival after adjusting for the predictions made by METSSS. Based on these discoveries, clinicians might choose to consult SORG-MLA instead of METSSS for survival estimation in patients with long-bone metastases presenting for evaluation of local treatment. Basing a treatment decision on the predictions of SORG-MLA could be beneficial when a patient's individualized risk threshold for opting to undergo a particular treatment strategy ranged from 0.2 to 0.9. Future studies might investigate relevant laboratory items when constructing or refining a survival estimation model because these data demonstrated prognostic value independent of the predictions of the METSSS model, and future studies might also seek to keep these models up to date using data from diverse, contemporary patients undergoing both modern operative and nonoperative treatments. Level III, diagnostic study.

Lee CC ，Chen CW ，Yen HK ，Lin YP ，Lai CY ，Wang JL ，Groot OQ ，Janssen SJ ，Schwab JH ，Hsu FM ，Lin WH ... -  《-》

Impact of residual disease as a prognostic factor for survival in women with advanced epithelial ovarian cancer after primary surgery.

Ovarian cancer is the seventh most common cancer among women and a leading cause of death from gynaecological malignancies. Epithelial ovarian cancer is the most common type, accounting for around 90% of all ovarian cancers. This specific type of ovarian cancer starts in the surface layer covering the ovary or lining of the fallopian tube. Surgery is performed either before chemotherapy (upfront or primary debulking surgery (PDS)) or in the middle of a course of treatment with chemotherapy (neoadjuvant chemotherapy (NACT) and interval debulking surgery (IDS)), with the aim of removing all visible tumour and achieving no macroscopic residual disease (NMRD). The aim of this review is to investigate the prognostic impact of size of residual disease nodules (RD) in women who received upfront or interval cytoreductive surgery for advanced (stage III and IV) epithelial ovarian cancer (EOC). To assess the prognostic impact of residual disease after primary surgery on survival outcomes for advanced (stage III and IV) epithelial ovarian cancer. In separate analyses, primary surgery included both upfront primary debulking surgery (PDS) followed by adjuvant chemotherapy and neoadjuvant chemotherapy followed by interval debulking surgery (IDS). Each residual disease threshold is considered as a separate prognostic factor. We searched CENTRAL (2021, Issue 8), MEDLINE via Ovid (to 30 August 2021) and Embase via Ovid (to 30 August 2021). We included survival data from studies of at least 100 women with advanced EOC after primary surgery. Residual disease was assessed as a prognostic factor in multivariate prognostic models. We excluded studies that reported fewer than 100 women, women with concurrent malignancies or studies that only reported unadjusted results. Women were included into two distinct groups: those who received PDS followed by platinum-based chemotherapy and those who received IDS, analysed separately. We included studies that reported all RD thresholds after surgery, but the main thresholds of interest were microscopic RD (labelled NMRD), RD 0.1 cm to 1 cm (small-volume residual disease (SVRD)) and RD > 1 cm (large-volume residual disease (LVRD)). Two review authors independently abstracted data and assessed risk of bias. Where possible, we synthesised the data in meta-analysis. To assess the adequacy of adjustment factors used in multivariate Cox models, we used the 'adjustment for other prognostic factors' and 'statistical analysis and reporting' domains of the quality in prognosis studies (QUIPS) tool. We also made judgements about the certainty of the evidence for each outcome in the main comparisons, using GRADE. We examined differences between FIGO stages III and IV for different thresholds of RD after primary surgery. We considered factors such as age, grade, length of follow-up, type and experience of surgeon, and type of surgery in the interpretation of any heterogeneity. We also performed sensitivity analyses that distinguished between studies that included NMRD in RD categories of < 1 cm and those that did not. This was applicable to comparisons involving RD < 1 cm with the exception of RD < 1 cm versus NMRD. We evaluated women undergoing PDS and IDS in separate analyses. We found 46 studies reporting multivariate prognostic analyses, including RD as a prognostic factor, which met our inclusion criteria: 22,376 women who underwent PDS and 3697 who underwent IDS, all with varying levels of RD. While we identified a range of different RD thresholds, we mainly report on comparisons that are the focus of a key area of clinical uncertainty (involving NMRD, SVRD and LVRD). The comparison involving any visible disease (RD > 0 cm) and NMRD was also important. SVRD versus NMRD in a PDS setting In PDS studies, most showed an increased risk of death in all RD groups when those with macroscopic RD (MRD) were compared to NMRD. Women who had SVRD after PDS had more than twice the risk of death compared to women with NMRD (hazard ratio (HR) 2.03, 95% confidence interval (CI) 1.80 to 2.29; I2 = 50%; 17 studies; 9404 participants; moderate-certainty). The analysis of progression-free survival found that women who had SVRD after PDS had nearly twice the risk of death compared to women with NMRD (HR 1.88, 95% CI 1.63 to 2.16; I2 = 63%; 10 studies; 6596 participants; moderate-certainty). LVRD versus SVRD in a PDS setting When we compared LVRD versus SVRD following surgery, the estimates were attenuated compared to NMRD comparisons. All analyses showed an overall survival benefit in women who had RD < 1 cm after surgery (HR 1.22, 95% CI 1.13 to 1.32; I2 = 0%; 5 studies; 6000 participants; moderate-certainty). The results were robust to analyses of progression-free survival. SVRD and LVRD versus NMRD in an IDS setting The one study that defined the categories as NMRD, SVRD and LVRD showed that women who had SVRD and LVRD after IDS had more than twice the risk of death compared to women who had NMRD (HR 2.09, 95% CI 1.20 to 3.66; 310 participants; I2 = 56%, and HR 2.23, 95% CI 1.49 to 3.34; 343 participants; I2 = 35%; very low-certainty, for SVRD versus NMRD and LVRD versus NMRD, respectively). LVRD versus SVRD + NMRD in an IDS setting Meta-analysis found that women who had LVRD had a greater risk of death and disease progression compared to women who had either SVRD or NMRD (HR 1.60, 95% CI 1.21 to 2.11; 6 studies; 1572 participants; I2 = 58% for overall survival and HR 1.76, 95% CI 1.23 to 2.52; 1145 participants; I2 = 60% for progression-free survival; very low-certainty). However, this result is biased as in all but one study it was not possible to distinguish NMRD within the < 1 cm thresholds. Only one study separated NMRD from SVRD; all others included NMRD in the SVRD group, which may create bias when comparing with LVRD, making interpretation challenging. MRD versus NMRD in an IDS setting Women who had any amount of MRD after IDS had more than twice the risk of death compared to women with NMRD (HR 2.11, 95% CI 1.35 to 3.29, I2 = 81%; 906 participants; very low-certainty). In a PDS setting, there is moderate-certainty evidence that the amount of RD after primary surgery is a prognostic factor for overall and progression-free survival in women with advanced ovarian cancer. We separated our analysis into three distinct categories for the survival outcome including NMRD, SVRD and LVRD. After IDS, there may be only two categories required, although this is based on very low-certainty evidence, as all but one study included NMRD in the SVRD category. The one study that separated NMRD from SVRD showed no improved survival outcome in the SVRD category, compared to LVRD. Further low-certainty evidence also supported restricting to two categories, where women who had any amount of MRD after IDS had a significantly greater risk of death compared to women with NMRD. Therefore, the evidence presented in this review cannot conclude that using three categories applies in an IDS setting (very low-certainty evidence), as was supported for PDS (which has convincing moderate-certainty evidence).

Bryant A ，Hiu S ，Kunonga PT ，Gajjar K ，Craig D ，Vale L ，Winter-Roach BA ，Elattar A ，Naik R ... -  《Cochrane Database of Systematic Reviews》

Application of a deep-learning marker for morbidity and mortality prediction derived from retinal photographs: a cohort development and validation study.

Biological ageing markers are useful to risk stratify morbidity and mortality more precisely than chronological age. In this study, we aimed to develop a novel deep-learning-based biological ageing marker (referred to as RetiPhenoAge hereafter) using retinal images and PhenoAge, a composite biomarker of phenotypic age. We used retinal photographs from the UK Biobank dataset to train a deep-learning algorithm to predict the composite score of PhenoAge. We used a deep convolutional neural network architecture with multiple layers to develop our deep-learning-based biological ageing marker, as RetiPhenoAge, with the aim of identifying patterns and features in the retina associated with variations of blood biomarkers related to renal, immune, liver functions, inflammation, and energy metabolism, and chronological age. We determined the performance of this biological ageing marker for the prediction of morbidity (cardiovascular disease and cancer events) and mortality (all-cause, cardiovascular disease, and cancer) in three independent cohorts (UK Biobank, the Singapore Epidemiology of Eye Diseases [SEED], and the Age-Related Eye Disease Study [AREDS] from the USA). We also compared the performance of RetiPhenoAge with two other known ageing biomarkers (hand grip strength and adjusted leukocyte telomere length) and one lifestyle factor (physical activity) for risk stratification of mortality and morbidity. We explored the underlying biology of RetiPhenoAge by assessing its associations with different systemic characteristics (eg, diabetes or hypertension) and blood metabolite levels. We also did a genome-wide association study to identify genetic variants associated with RetiPhenoAge, followed by expression quantitative trait loci mapping, a gene-based analysis, and a gene-set analysis. Cox proportional hazards models were used to estimate the hazard ratios (HRs) and corresponding 95% CIs for the associations between RetiPhenoAge and the different morbidity and mortality outcomes. Retinal photographs for 34 061 UK Biobank participants were used to train the model, and data for 9429 participants from the SEED cohort and for 3986 participants from the AREDS cohort were included in the study. RetiPhenoAge was associated with all-cause mortality (HR 1·92 [95% CI 1·42-2·61]), cardiovascular disease mortality (1·97 [1·02-3·82]), cancer mortality (2·07 [1·29-3·33]), and cardiovascular disease events (1·70 [1·17-2·47]), independent of PhenoAge and other possible confounders. Similar findings were found in the two independent cohorts (HR 1·67 [1·21-2·31] for cardiovascular disease mortality in SEED and 2·07 [1·10-3·92] in AREDS). RetiPhenoAge had stronger associations with mortality and morbidity than did hand grip strength, telomere length, and physical activity. We identified two genetic variants that were significantly associated with RetiPhenoAge (single nucleotide polymorphisms rs3791224 and rs8001273), and were linked to expression quantitative trait locis in various tissues, including the heart, kidneys, and the brain. Our new deep-learning-derived biological ageing marker is a robust predictor of mortality and morbidity outcomes and could be used as a novel non-invasive method to measure ageing. Singapore National Medical Research Council and Agency for Science, Technology and Research, Singapore.

Nusinovici S ，Rim TH ，Li H ，Yu M ，Deshmukh M ，Quek TC ，Lee G ，Chong CCY ，Peng Q ，Xue CC ，Zhu Z ，Chew EY ，Sabanayagam C ，Wong TY ，Tham YC ，Cheng CY ... -  《The Lancet Healthy Longevity》

Effectiveness of interventions for improving educational outcomes for people with disabilities in low- and middle-income countries: A systematic review.

People with disabilities are consistently falling behind in educational outcomes compared to their peers without disabilities, whether measured in terms of school enrolment, school completion, mean years of schooling, or literacy levels. These inequalities in education contribute to people with disabilities being less likely to achieve employment, or earn as much if they are employed, as people without disabilities. Evidence suggests that the gap in educational attainment for people with and without disabilities is greatest in low- and middle-income countries (LMICs). Exclusion of people with disabilities from mainstream education, and low rates of participation in education of any kind, are important issues for global equity. Interventions which might have a positive impact include those that improve educational outcomes for people with disabilities, whether delivered in specialist or inclusive education settings. Such interventions involve a wide range of initiatives, from those focused on the individual level - such as teaching assistance to make mainstream classes more accessible to children with specific learning needs - to those which address policy or advocacy. The objectives of this review were to answer the following research questions: (1) What is the nature of the interventions used to support education for people with disabilities in LMICs? (2) What is the size and quality of the evidence base of the effectiveness of interventions to improve educational outcomes for people with disabilities in LMICs? (3) What works to improve educational outcomes for people with disabilities in LMICs? (4) Which interventions appear to be most effective for different types of disability? (5) What are the barriers and facilitators to the improvement of educational outcomes for people with disabilities? (6) Is there evidence of cumulative effects of interventions? The search for studies followed two steps. Firstly, we conducted an electronic search of databases and sector-specific websites. Then, after initial screening, we examined the reference lists of all identified reviews and screened the cited studies for inclusion. We also conducted a forward search and an ancestral search. No restrictions in terms of date or format were placed on the search, but only English-language publications were eligible for inclusion. In our review, we included studies on the basis that they were able to detect intervention impact. Descriptive studies of various designs and methodologies were not included. We also excluded any study with a sample size of fewer than five participants. We included studies which examined the impact of interventions for people with disabilities living in LMICs. There were no restrictions on comparators/comparison groups in included studies. However, to be eligible for inclusion, a study needed to have both an eligible intervention and an eligible outcome. Any duration of follow-up was eligible for inclusion. We used EppiReviewer for bibliographic management, screening, coding, and data synthesis. Eligibility was assessed using a predesigned form based on the inclusion criteria developed by the authors. We piloted all coding sheets with at least five studies before use. The form allowed for coding of multiple intervention domains and multiple outcomes domains. The entire screening process was reported using a PRISMA flow chart. We screened all unique references from our search title and abstract, with two independent reviewers determining relevance, and repeated this process for full texts. Data was extracted from studies according to a coding sheet. Coding included: (1) extraction of basic study characteristics, (2) a narrative summary of procedures and findings (including recording of iatrogenic effects), (3) a summary of findings/results table, (4) an assessment of confidence in study findings, and (5) creation of a forest plot of effect sizes. A third data collector, a research associate, checked the results of this process. Confidence in study findings was assessed using a standardised tool. All coding categories were not mutually exclusive and so multiple coding was done where an intervention covered more than one category of intervention. Twenty-eight studies were included in this review. Most studies (n = 25) targeted children with disabilities. Only two studies directly targeted family members, and the remaining three focused on service providers. Individuals with intellectual or learning and developmental impairments were most frequently targeted by interventions (n = 17). The category of interventions most represented across studies was 'Educational attainment support', for instance, a reading comprehension intervention that combined strategy instruction (graphic organisers, visual displays, mnemonic illustrations, computer exercises, predicting, inference, text structure awareness, main idea identification, summarisation, and questioning) for children with dyslexia. The second most common category of intervention was 'Accessible learning environments', for instance, programmes which aimed to improve social skills or to reduce rates of victimisation of children with disabilities in schools. Regarding intervention effects, included studies concerned with 'Conditions for inclusion of people with disabilities in education' showed a moderately significant effect, and one study concerned with teacher knowledge showed a significant effect size. Among the 18 studies included in the analysis of intervention effects on 'Skills for learning', 12 interventions had a significant effect. When considering the effect of interventions on different outcomes, we see that the effect on literacy, cognitive skills, handwriting, and numeracy are significant. All these effects are large but are based on a low number of studies. The studies concerned with speech and school behaviour show no significant effect of intervention. Across studies, heterogeneity is high, and risk of publication bias varies but was frequently high. All but one study received an overall rating of low confidence in study findings. However, this lack of confidence across studies was largely due to the use of low-rigour study designs and was not always reflective of multiple points of weakness within a given study. Children with disabilities fall behind in educational outcomes as the current school systems are not set up to teach children with different impairment types. There is no one 'magic bullet' intervention which can equalise health outcomes for this group. A twin-track approach is needed, which both addresses the specific needs of children with disabilities but also ensures that they are included in mainstream activities (e.g., through improving the skills of teachers and accessibility of the classroom). However, currently most interventions included in this systematic review targeted individual children with disabilities in an attempt to improve their functioning, skills, and competencies, but did not focus on mainstreaming these children into the school by system-level or school-level changes. Consequently, a focus on evaluation of interventions which target not just the individual with a disability but also their broader environment, are needed.

Hunt X ，Saran A ，White H ，Kuper H ... -  《-》