Artificial Intelligence Platform on Mobile Devices to Assess Consumption of Pill in Subjects with Alzheimer

Abstract

There are many misconceptions in the medical community concerning the importance of drug monitoring. In the United States alone, between 7,000 and 9,000 people die each year because of medication errors. The total cost of treating patients with medication errors is over $40 billion (about $120 per person in the United States) every year, with more than 7 million patients (two times the population of Oklahoma) affected. Patients incur psychological and physical pain and suffering because of drug errors, in addition to the financial costs. Finally, pharmaceutical errors result in a decrease in patient satisfaction and a rising lack of trust in the healthcare system. Failure to convey medicine prescriptions, unreadable handwriting, poor drug selection from a drop-down option, confusion about drugs with similar names, confusion about similar packaging between products, and dosage or weight errors are among the most prevalent causes of errors. Human mistakes can cause medication errors, although they are more typically caused by a failing system with insufficient backup to detect errors. In the case of adverse medication responses, the patient is harmed by taking a drug as prescribed, does not have a necessary drug, or has received it incorrectly, such as at an excessively high or low dose. However, in recent years, medication surveillance has not only gained support across the health profession but has also been added to regulatory guidelines from the US Centers for Disease Control and Prevention CDC (Centers for Disease Control) and is accepted by various medical councils. check whether the patient has consumed the pill or not. In our application to monitor Pill consumption, we are using a Single Shot detector that takes a single exposure to detect several objects in an image using a multi-box. As a highlight, the resulting Single Shot Detector model attains 92% accuracy on MobileNet. The results of our Single Shot Detector Model attain 92% accuracy on MobileNet. This application will help to reduce the labor cost of hospitals.