This work presents the development of devices for remote monitoring of manic-depressive, elderly, and obese people to be used at IoT (Internet of Things) context. These people are monitored through a wearable device and an instrumented digital scale. For monitoring of manic-depressive and ageing people, several techniques were analyzed (Decision tree, kNearest Neighbor, Naive Bayes, and Artificial Neural Network) to allow human activity recognition. k-Nearest Neighbor got the best result (86% of accuracy) for four different trials in six different activities (walking, running, going upstairs, going downstairs, standing up, and sitting). For monitoring of obese people, an app was developed for a smartphone, which obtains remote weight measurements with 0.01 kg of accuracy.