Grading Acanthosis Nigricans Using a Smartphone and Color Analysis: A Novel Noninvasive Method to Screen for Impaired Glucose Tolerance and Type 2 Diabetes.

Objective: The objective of this study was to develop ANcam, a novel method for identifying acanthosis nigricans (AN) using a smartphone camera and computer-aided color analysis for noninvasive screening of people with impaired glucose tolerance (IGT). Research Design and Methods: Adult and juvenile participants with or without diagnosed type 2 diabetes were recruited in Trinidad and Tobago. After obtaining informed consent, participants' history, demographics, anthropometrics, and A1C were collected and recorded. Three subject matter experts independently graded pictures of the posterior neck and upper back using the ANcam smartphone application and Burke methods. A correlation matrix investigated 25 color channels for association with hyperpigmentation, and the diagnostic thresholds were determined with a receiver operating characteristic curve analysis. Results: For the 227 participants with captured images and A1C values, the cyan/magenta/yellow/black (CMYK) model color channel CMYK_K was best correlated with IGT at an A1C cutoff of 5.7% (39 mmol/mol) (R = 0.45, P <0.001). With high predictive accuracy (area under the curve = 0.854), the cutoff of 7.67 CMYK_K units was chosen, with a sensitivity of 81.1% and a specificity of 70.3%. ANcam had low interrater variance (F = 1.99, P = 0.137) compared with Burke grading (F = 105.71, P <0.001). ANcam detected hyperpigmentation on the neck at double the self-reported frequency. Elevated BMI was 2.9 (95% CI 1.9-4.3) times more likely, elevated blood pressure was 1.7 (95% CI 1.2-2.4) times more likely, and greater waist-to-hip ratio was 2.3 (95% CI 1.4-3.6) times more likely with AN present. Conclusion: ANcam offers a sensitive, reproducible, and user-friendly IGT screening tool to any smartphone user that performs well with most skin tones and lighting conditions.

Optimisation of cancer classification by machine learning generates an enriched list of candidate drug targets and biomarkers.

The Cancer Genome Atlas has provided expression values of 18 015 genes for different cancer types. Studies on the classification of cancers by machine learning algorithms have used different data and methods, which makes it difficult to compare their performance. It is unclear, which algorithm performs best and if maximum levels of accuracy have been obtained. In this study, we aimed to optimise the diagnosis of cancer by comparing the performance of five algorithms using the same data, and by identifying the smallest possible number of differentiator genes. Classification accuracies of five algorithms of cancer type and primary site were determined using a gene expression dataset of 5629 samples and a dataset of 9144 samples, respectively. When trained with sample sets ranging from 16 718 to 40 genes, Random Forest (RF), Gradient Boosting Machine (GBM), and Neural Network (NN) consistently achieved 100% or near 100% accuracy in the classification of both cancer type and primary site. Reduction of training sets to the 40 highest-ranked genes resulted in 78-fold and 45-fold faster processing times for RF and GBM, respectively. The olfactory receptor family, keratin associated proteins, and defensin beta family were among the highest ranked genes. The ensemble and NN algorithms were the most accurate at distinguishing between cancer types and primary sites, whereas KNN was the fastest. Training sets can be reduced to the 40 highest-ranked differentiator genes without any significant loss of accuracy, amongst which there are potential drug targets and biomarkers.