Importance Deep learning is a family of computational methods that allow an algorithm to program itself by learning from a large set of examples that demonstrate the desired behavior, removing the need to specify rules explicitly. Application of these methods to medical imaging requires further assessment and validation. Objective To apply deep learning to create an algorithm for automated detection of diabetic retinopathy and diabetic macular edema in retinal fundus photographs. Design and Setting A specific type of neural network optimized for image classification called a deep convolutional neural network was trained using a retrospective development data set of 128 175 retinal images, which were graded 3 to 7 times for diabetic retinopathy, diabetic macular edema, and image gradability by a panel of 54 US licensed ophthalmologists and ophthalmology senior residents between May and December 2015. The resultant algorithm was validated in January and February 2016 using 2 separate data sets, both graded by at least 7 US board-certified ophthalmologists with high intragrader consistency. Exposure Deep learning–trained algorithm. Main Outcomes and Measures The sensitivity and specificity of the algorithm for detecting referable diabetic retinopathy (RDR), defined as moderate and worse diabetic retinopathy, referable diabetic macular edema, or both, were generated based on the reference standard of the majority decision of the ophthalmologist panel. The algorithm was evaluated at 2 operating points selected from the development set, one selected for high specificity and another for high sensitivity. Results The EyePACS-1 data set consisted of 9963 images from 4997 patients (mean age, 54.4 years; 62.2% women; prevalence of RDR, 683/8878 fully gradable images [7.8%]); the Messidor-2 data set had 1748 images from 874 patients (mean age, 57.6 years; 42.6% women; prevalence of RDR, 254/1745 fully gradable images [14.6%]). For detecting RDR, the algorithm had an area under the receiver operating curve of 0.991 (95% CI, 0.988-0.993) for EyePACS-1 and 0.990 (95% CI, 0.986-0.995) for Messidor-2. Using the first operating cut point with high specificity, for EyePACS-1, the sensitivity was 90.3% (95% CI, 87.5%-92.7%) and the specificity was 98.1% (95% CI, 97.8%-98.5%). For Messidor-2, the sensitivity was 87.0% (95% CI, 81.1%-91.0%) and the specificity was 98.5% (95% CI, 97.7%-99.1%). Using a second operating point with high sensitivity in the development set, for EyePACS-1 the sensitivity was 97.5% and specificity was 93.4% and for Messidor-2 the sensitivity was 96.1% and specificity was 93.9%. Conclusions and Relevance In this evaluation of retinal fundus photographs from adults with diabetes, an algorithm based on deep machine learning had high sensitivity and specificity for detecting referable diabetic retinopathy. Further research is necessary to determine the feasibility of applying this algorithm in the clinical setting and to determine whether use of the algorithm could lead to improved care and outcomes compared with current ophthalmologic assessment.

/pdf/development-and-validation-of-a-deep-learning-algorithm-for-4ql9aa9p1e.pdf

Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs

The following CDC staff members prepared this report: National Center for Infectious Diseases Division of HIV/AIDS Kenneth G. Castro, M.D. John W. Ward, M.D. Laurence Slutsker, M.D., M.P.H. James W. Buehler, M.D. Harold W. Jaffe, M.D. Ruth L. Berkelman, M.D. Office of the Director Associate Director for HIV/AIDS James W. Curran, M.D., M.P.H. 1993 Revised Classification System for HIV Infection and Expanded Surveillance Case Definition for AIDS Among Adolescents and Adults Summary CDC has revised the classification system for HIV infection to emphasize the clinical importance of the CD4+ T-lymphocyte count in the categorization of HIV-related clinical conditions. This classification system replaces the system published by CDC in 1986 (1) and is primarily intended for use in public health practice. Consistent with the 1993 revised classification system, CDC has also expanded the AIDS surveillance case definition to include all HIV-infected persons who have less than 200 CD4+ T-lymphocytes/uL, or a CD4+ T-lymphocyte percentage of total lymphocytes of less than 14. This expansion includes the addition of three clinical conditions

/pdf/1993-revised-classification-system-for-hiv-infection-and-4ldnyqo2jk.pdf

1993 Revised Classification System for HIV Infection and Expanded Surveillance Case Definition for AIDS Among Adolescents and Adults

background Chronic obstructive pulmonary disease (COPD) is characterized by an incompletely reversible limitation in airflow. A physiological variable — the forced expiratory volume in one second (FEV 1 ) — is often used to grade the severity of COPD. However, patients with COPD have systemic manifestations that are not reflected by the FEV 1 . We hypothesized that a multidimensional grading system that assessed the respiratory and systemic expressions of COPD would better categorize and predict outcome in these patients. methods We first evaluated 207 patients and found that four factors predicted the risk of death in this cohort: the body-mass index (B), the degree of airflow obstruction (O) and dyspnea (D), and exercise capacity (E), measured by the six-minute–walk test. We used these variables to construct the BODE index, a multidimensional 10-point scale in which higher scores indicate a higher risk of death. We then prospectively validated the index in a cohort of 625 patients, with death from any cause and from respiratory causes as the outcome variables. results There were 25 deaths among the first 207 patients and 162 deaths (26 percent) in the validation cohort. Sixty-one percent of the deaths in the validation cohort were due to respiratory insufficiency, 14 percent to myocardial infarction, 12 percent to lung cancer, and 13 percent to other causes. Patients with higher BODE scores were at higher risk for death; the hazard ratio for death from any cause per one-point increase in the BODE score was 1.34 (95 percent confidence interval, 1.26 to 1.42; P<0.001), and the hazard ratio for death from respiratory causes was 1.62 (95 percent confidence interval, 1.48 to 1.77; P<0.001). The C statistic for the ability of the BODE index to predict the risk of death was larger than that for the FEV 1 (0.74 vs. 0.65).

https://www.taiss.com/gepoc/gecarp/gepubli/publdoc/026-026-cellibr-body-mass-index-airflow-obstruction-dyspnea.pdf

The Body-Mass Index, Airflow Obstruction, Dyspnea, and Exercise Capacity Index in Chronic Obstructive Pulmonary Disease

HEALTH-related quality of life (HRQL) is increasingly used as an outcome in clinical trials, effectiveness research, and research on quality of care. Factors that have facilitated this increased usage include the accumulating evidence that measures of HRQL are valid and "reliable,"1the publication of several large clinical trials showing that these outcome measures are responsive to important clinical changes,2-5and the successful development and testing of shorter instruments that are easier to understand and administer.6-13Because these measures describe or characterize what the patient has experienced as the result of medical care, they are useful and important supplements to traditional physiological or biological measures of health status. Given this improved ability to assess patients' health status, how can physicians and health care systems intervene to improve HRQL? Implicit in the use of measures of HRQL in clinical trials and in effectiveness research is the concept that clinical

Linking clinical variables with health-related quality of life. A conceptual model of patient outcomes

SUMMARY OF APPROACH The approach to surgery in the patient with COPD is summa-rized below. Surgery Definitely indicated Lung resection. Pulmonary function studies should be performedbefore lung resection. Simple spirometry has the greatest utilityin documenting physiologic operability. FEV, > 2 L in an adultman or > 60% of predicted is acceptable for pneumonectomy.Values below this suggest that further studies, such as split func-tion assessment by quantitative lung scintigraphy and exercisetesting, are warranted.FEV, predicted after lung resection to be less than 40 to 50%of normal for the patient’s sex, age, and height suggests highermorbidity and mortality. An exercise 90, of less than 10 to 15ml/min per kg of body weight is associated with higher mor-bidity and mortality after lung resection.All elective surgery Prophylaxis against deep venous throm-bosis should be given before most procedures that will requirepostoperative bed rest or significantly reduce mobility. Heparinin low doses seems well accepted for most procedures. Externalpneumatic compression of the lower legs can be used when anti-coagulants are contraindicated.

/pdf/standards-for-the-diagnosis-and-care-of-patients-with-13vz6fyao0.pdf

Standards for the Diagnosis and Care of Patients with Chronic Obstructive Pulmonary Disease

We found that a cohort of patients with lung cancer first treated in 1977 had higher six-month survival rates for the total group and for subgroups in each of the three main TNM stages (tumor, nodes, and metastases) than a cohort treated between 1953 and 1964 at the same institutions. The more recent cohort, however, had undergone many new diagnostic imaging procedures. According to the "old" diagnostic data for both cohorts, the recent cohort had a prognostically favorable "zero-time shift." In addition, by demonstrating metastases that had formerly been silent and unidentified, the new technological data resulted in a stage migration. Many patients who previously would have been classified in a "good" stage were assigned to a "bad" stage. Because the prognosis of those who migrated, although worse than that for other members of the good-stage group, was better than that for other members of the bad-stage group, survival rates rose in each group without any change in individual outcomes. When cl...

The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer.

Evaluation of Clinical Methods for Rating Dyspnea

https://journal.chestnet.org/article/S0012-3692(16)62410-8/pdf

The measurement of dyspnea. Contents, interobserver agreement, and physiologic correlates of two new clinical indexes.

Background Despite the proved value of mammography in screening for breast cancer, its efficacy depends on radiologists' interpretations. The variability in such interpretations is not well understood. Methods Using a technique of stratified random sampling, we selected 150 mammograms obtained in 1987: 27 from women with histopathologically confirmed breast cancer and 123 from women with no evidence of breast cancer after three years of follow-up examinations. Ten radiologists, who were unaware of the diagnoses and research hypothesis, each interpreted the 150 mammograms. Disagreement was analyzed within pairs of the 10 radiologists, as well as for the group of 150 women as a whole. Results The diagnostic consistency between pairs of radiologists was moderate, with a median weighted percentage of agreement of 78 percent (weighted kappa, 0.47). The frequency of the radiologists' recommendations for an immediate workup ranged from 74 to 96 percent for mammograms from the women with cancer and from 11 to 65 ...

https://www.nejm.org/doi/pdf/10.1056/NEJM199412013312206

Variability in Radiologists' Interpretations of Mammograms

Indexes of functional disability are being used increasingly to rate the status of patients studied in clinical research or treated in clinical practice. To determine why so many indexes h...

Carolyn K. Wells

Papers

The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer.

Evaluation of Clinical Methods for Rating Dyspnea

The measurement of dyspnea. Contents, interobserver agreement, and physiologic correlates of two new clinical indexes.

Variability in Radiologists' Interpretations of Mammograms

Scientific and clinical problems in indexes of functional disability.