Heart disease (HD) is the leading cause of death worldwide. The early and accurate diagnosis of HD is the key to reduce mortality rate. To diagnose HD, vital signs such as blood sugar, heart rate, electrocardiogram (ECG), blood pressure, cholesterol level, etc. are needed. These biomarkers are generally obtained from sensors, such as pulse sensor, blood pressure sensor, etc., attached to a person's body. Integrating internet of things (IoT), artificial intelligence (AI), and cloud computing technologies automates the HD diagnosis process. IoT devices are used to collect and transmit patient physiological data to an AI-based HD diagnosis model deployed in a cloud computing platform. Many studies developed IoT-based HD diagnostic systems using deep learning (DL) models, a branch of AI. However, most focused on improving diagnostic accuracy by increasing DL model complexity, which leads to more computation time and need for larger training data. In this thesis, different DL models are proposed, developed, and implemented for HD diagnosis. Cleveland, Statlog, and Comprehensive benchmark datasets from UCI are utilized to train and test the models. Traditional approaches including support vector machine (SVM) and artificial neural network (ANN), were used to diagnose HD. Each achieved a satisfactory accuracy level (89.7% and 93.44% respectively). Two simplified DL models based on one-dimensional convolutional neural network (1D-CNN) and bi-directional long short-term memory (Bi-LSTM) were proposed to improve diagnosis accuracy. Both reached 94.96% accuracy. This accuracy could not further enhance due to the small and imbalanced dataset. To overcome this issue, DL-based data generative model employing generative adversarial network (GAN) was developed. The GAN model generates additional data to enlarge the small and imbalanced dataset. Then, the 1D-CNN and Bi-LSTM models are retrained using the enlarger dataset. This boosted 1D-CNN accuracy to 99.1% and Bi-LSTM accuracy to 99.3%. Moreover, principal component analysis (PCA) is used to investigate the time complexity of the proposed 1D-CNN and Bi-LSTM models. PCA reduced the dataset dimensionality which lowered the 1D-CNN and Bi-LSTM prediction times to 68.8 and 74.8 ms, respectively. Finally, ANN, 1D-CNN, and Bi-LSTM models are stacked in an ensemble model to reduce the diagnosis error of each one. The stacked ensemble model improved the overall diagnosis accuracy to 99.7%. Consequently, this model was deployed on the Amazon cloud computing platform to diagnose HD from sensors data and display the diagnosis result on Android application in a complete IoT environment.

A stroke can cause upper limb (UL) dysfunction that limits movement. Consequently, the patient cannot practice daily life activities normally, recovery after a stroke is necessary, and the patient must undergo rehabilitation to restore UL function as much as possible. Active rehabilitation techniques can hasten a motor system recovery but with significant time, like conventional rehabilitation that includes the conventional intervention (CI), which uses traditional methods in the recovery process. Virtual reality (VR) rehabilitation that includes VR intervention (VRI) which employs VR technology, is an attractive method that shows promising results for stroke recovery that can improve the effectiveness of current rehabilitation procedures in terms of the quality of recovery methods and the ability to obtain satisfactory results in a short time. The work of this thesis was conducted at the Al-Hamza Specialized Center for stroke patients’ Rehabilitation in Baghdad/Iraq, where 30 stroke patients participated in the evaluation of the proposed system. All patients underwent CI that used the real game of grasping and throwing five balls through three levels. After a rest period, they underwent VRI that used the same game but in a VR environment. Times for grasping and throwing the balls were measured for the real and VR games, and then, these two times were compared for conventional and VR techniques to find the effectiveness of the proposed VR rehabilitation system All patients had their disability assessed using the stroke assessment scales, which were the Fugl-Meyer Assessment for Upper Extremity (FMA-UE) and Wolf Motor Function Test (WMFT). These assessment scales were used before and after the experiment (playing the games) to measure how much they had benefited from them. The goal of this thesis is to present a VR-based rehabilitation system, that includes building a VR game with three gradual difficulty levels to suit all cases of disability in stroke patients, The proposed system gives the therapist the ability to monitor patient interactions in the real and virtual world simultaneously. The result showed that the dysfunction UL for the participant patients was improved by using VRI effectively compared with CI; for conventional rehabilitation, the gain in FMA-UE and WMFT before and after the experiment was (0.65) points for FMA-UE and (1.485) seconds for WMFT. When using the VR rehabilitation, the gain in FMA-UE was (3.75) points, and the gain for WMFT was (7.637) seconds. Patients were able to reduce the times of grasping (TG) and the times of throwing (TT)in the VRI by about five times the CI. Depending on the results of the FMA-UE and WMFT and the times of TG and TT, the medical staff and specialists of the Al-Hamza rehabilitation center which includes the doctors who specialize in rehabilitation confirmed that VRI was more valuable and effective in improving the upper motor system of the patients after the stroke. Medical rehabilitation centers can utilize the work of this thesis to rehabilitate stroke patients in Iraq, especially as they lack such modern technology. The use of VR has spread in all hospitals and medical centers worldwide. Hence the proposed system is considered an opportunity for patients to improve their disability in the ULs by using the newest technology in the world, which can save effort and time for the patient through the recovery process compared with conventional techniques