Personal Artificial Intelligence Doctor Our solution utilizes Machine Learning (ML) algorithms in order to accomplish its goals. The main ML concept used is the Natural Language Processing (NLP), specifically word embedding word2vec method. The idea is to build a word2vec model using medical text such as academic books, clinics archive, and web content such as Wikipedia and other medical oriented websites. Such matrix will contain a significant amount of knowledge regarding symptoms and diseases, however unstructured, therefor, this word2vec model will be used as keystone to constructing a three clusters, symptoms cluster, diseases cluster, and best practices cluster, Afterwords, a sim-supervised approach consisting of manually creating the clusters with a specialist doctor help then update the clusters to the full amount of available information using the word2vec model. However our solution have more functions using ML such as audio and image classification, simply the user will record there voice or capture an image using the mobile app and the solution will classify the data as healthy or not health using Extreme Learning Machine (ELM) for audio, and Convolutional Neural Network CNN for image classification.
PILOT PSAIDR

أولا : يقوم المريض بادخال الأعراض التي يشعر بها سواء باللهجة العامية او الفصحى

ثانيا : يقوم البرنامج بادخال الكلام الى الذكاء الاصطناعي الذي يفحص الكلمات ويفهمها مثلا : لو كتب ( منخاري يوجعني ) يرجع
الذكاء الاصطناعي جملة ( ألم بالأنف ) .

ثالثا : فلنفترض أن المريض قام بإدخال جملة : ( أشعر بالدوار والغثيان والصداع ) يقوم الذكاء الاصطناعي بسؤال 5 اسئلة لها
علاقة بهذه الأعراض ليجيب المريض عن سؤال واحد ويتم تحديد أقرب مرض

رابعا : يسأل البرنامج 3 اسئلة تكون من اعراض الامراض التي تحتوي على هذه الأعراض

خامسا : يجيب المريض على الاسئلة وبحسب اجاباته سيتم تحديد أي مرض له أعراض أكثر يشعر بها فيتم تحديد المرض, و تتكرر
الخطوة الرابعة و الخامسة عند الحاجة حتى يتم تشخيص المرض بشكل صحيح.

PILOT PSAIDR

First: The patient enters the symptoms he or she feels in either colloquial or standard language
Second: The program feed the words to Artificial Intelligence, which examines words and
understands them. For example, if written (Mnkhari Yogani ) the intelligence returns the
sentence (pain in the nose) Mnkhari (منخاري): is a slang word for nose, commonly used in palestinan tongue.
Third: Suppose that the patient entered the sentence: (I feel dizzy, nausea and headache)
Artificial Intelligence ask 5 questions related to these symptoms. Then patient selects one question
to determine the nearest symptom
Fourth : Then the AI asks 3 questions ( symptoms ) from the diseases that contain these symptoms
to shortlist and diagnose the disease.

Fifth: Then the patient answers these questions, and according to the answers the AI will determine
which disease has more symptoms in common with the user input, process four and five are
repeated until AI diagnosis is completed successfully .