diff --git a/README.md b/README.md index 8bd7aaea0..cf912602b 100644 --- a/README.md +++ b/README.md @@ -137,11 +137,6 @@ Convolutional Neural Network and Image processing of 136,756 mammogram images we * Risk Factor Dataset can be changed for a more general dataset that includes women and men. It would be ideal to conduct an analysis of a longitudinal data set to closely observe the transformation of some of the features and how they may lead to cancer diagnosis. -* CNN Machine Learning Model should have a greater positive predictive accuracy score. Because this has a medical application, we should aspite to have above 98% 99% accuracy for TRUE POTIVES. Model optimization can be done through better feature selection, increasing training and testing images, and/or adding more hidden layers to the model architecture. We plan to keep improving the model. +* By leveraging CNN Machine Learning Model, we can produce better predictive accuracy score. Since this can be applied in the field of medicine, targeting higher than 98% accuracy for positive findings is necessary to make "life-saving" impact. Model optimization can be achieved through better feature selection, increasing training, different test images, and adding more hidden layers to the model architecture. Improving the model is a crucial first step. -* Web app can be used as a screening tool at medical center to direct patients to specialist. We need to adapt the code to allow users to upload mammograms. - -## Communication and More Information ## - -We are currently communicating through a Slack channel (name: #finalproject); and have created a Google Drive Folder to store presentation templetes and meeting minutes. -[link](https://drive.google.com/drive/folders/1LaX9KnGGwv2Lf1l9v83TqIOjVzP_935L) +* Web apps can be used as a screening tool at medical center first before directing patients to specialist. By giving users capabilities to upload mammograms and self-serve is highly recommended.