Assessment of Grain Harvest Moisture Content Using Machine Learning on Smartphone Images for Optimal Harvest Timing.

Grain moisture content (GMC) is a key indicator of the appropriate harvest period of rice. Conventional testing is time-consuming and laborious, thus not to be implemented over vast areas and to enable the estimation of future changes for revealing optimal harvesting. Images of single panicles were shot with smartphones and corrected using a spectral-geometric correction board. In total, 86 panicle samples were obtained each time and then dried at 80 °C for 7 days to acquire the wet-basis GMC. In total, 517 valid samples were obtained, in which 80% was randomly used for training and 20% was used for testing to construct the image-based GMC assessment model. In total, 17 GMC surveys from a total of 201 samples were also performed from an area of 1 m2 representing on-site GMC, which enabled a multi-day GMC prediction. Eight color indices were selected using principal component analysis for building four machine learning models, including random forest, multilayer perceptron, support vector regression (SVR), and multivariate linear regression. The SVR model with a MAE of 1.23% was the most suitable for GMC of less than 40%. This study provides a real-time and cost-effective non-destructive GMC measurement using smartphones that enables on-farm prediction of harvest dates and facilitates the harvesting scheduling of agricultural machinery.

Gene expression profile of peripheral blood in colorectal cancer.

AIM: Optimal molecular markers for detecting colorectal cancer (CRC) in a blood-based assay were evaluated. METHODS: A matched (by variables of age and sex) case-control design (111 CRC and 227 non-cancer samples) was applied. Total RNAs isolated from the 338 blood samples were reverse-transcribed, and the relative transcript levels of candidate genes were analyzed. The training set was made of 162 random samples of the total 338 samples. A logistic regression analysis was performed, and odds ratios for each gene were determined between CRC and non-cancer. The samples (n = 176) in the testing set were used to validate the logistic model, and an inferred performance (generality) was verified. By pooling 12 public microarray datasets(GSE 4107, 4183, 8671, 9348, 10961, 13067, 13294, 13471, 14333, 15960, 17538, and 18105), which included 519 cases of adenocarcinoma and 88 controls of normal mucosa, we were able to verify the selected genes from logistic models and estimate their external generality. RESULTS: The logistic regression analysis resulted in the selection of five significant genes (P < 0.05; MDM2, DUSP6, CPEB4, MMD, and EIF2S3), with odds ratios of 2.978, 6.029, 3.776, 0.538 and 0.138, respectively. The five-gene model performed stably for the discrimination of CRC cases from controls in the training set, with accuracies ranging from 73.9% to 87.0%, a sensitivity of 95% and a specificity of 95%. In addition, a good performance in the test set was obtained using the discrimination model, providing 83.5% accuracy, 66.0% sensitivity, 92.0% specificity, a positive predictive value of 89.2% and a negative predictive value of 73.0%. Multivariate logistic regressions analyzed 12 pooled public microarray data sets as an external validation. Models that provided similar expected and observed event rates in subgroups were termed well calibrated. A model in which MDM2, DUSP6, CPEB4, MMD, and EIF2S3 were selected showed the result in logistic regression analysis (H-L P = 0.460, R2= 0.853, AUC = 0.978, accuracy = 0.949, specificity = 0.818 and sensitivity = 0.971). CONCLUSION: A novel gene expression profile was associated with CRC and can potentially be applied to blood-based detection assays.