Source-linked AI summary
A Novel AI-enabled Framework to Diagnose Coronavirus COVID 19 using Smartphone Embedded Sensors: Design Study
Halgurd S. Maghdid, Kayhan Zrar Ghafoor, Ali Safaa Sadiq, Kevin Curran, Danda B. Rawat, Khaled Rabie
TL;DR
The paper addresses costly, time-consuming, and overloaded COVID-19 diagnosis workflows. It proposes an AI-enabled smartphone framework that combines built-in sensor measurements with CT-image analysis to predict disease and pneumonia severity, and reports a low-cost, accessible design intended for radiologists and smartphone users.
Problem
COVID-19 diagnosis can involve costly or imperfect methods, while demand for CT examinations overloads medical systems and contributes to delayed detection and treatment.
Method
The framework combines built-in smartphone sensors, symptom-processing algorithms, CT-image analysis, machine learning, and cloud-based data exchange.
Results
The paper presents a proof-of-concept framework and mobile app that capture smartphone sensor data and CT images for COVID-19 diagnosis and pneumonia-severity prediction.
Takeaways & Limitations
The proposed design is intended to provide a low-cost smartphone-based option usable by radiologists and ordinary people for diagnosis and inflammation monitoring.
Abstract
from arXiv · showhide
Coronaviruses are a famous family of viruses that cause illness in both humans and animals. The new type of coronavirus COVID-19 was firstly discovered in Wuhan, China. However, recently, the virus has widely spread in most of the world and causing a pandemic according to the World Health Organization (WHO). Further, nowadays, all the world countries are striving to control the COVID-19. There are many mechanisms to detect coronavirus including clinical analysis of chest CT scan images and blood test results. The confirmed COVID-19 patient manifests as fever, tiredness, and dry cough. Particularly, several techniques can be used to detect the initial results of the virus such as medical detection Kits. However, such devices are incurring huge cost, taking time to install them and use. Therefore, in this paper, a new framework is proposed to detect COVID-19 using built-in smartphone sensors. The proposal provides a low-cost solution, since most of radiologists have already held smartphones for different daily-purposes. Not only that but also ordinary people can use the framework on their smartphones for the virus detection purposes. Nowadays Smartphones are powerful with existing computation-rich processors, memory space, and large number of sensors including cameras, microphone, temperature sensor, inertial sensors, proximity, colour-sensor, humidity-sensor, and wireless chipsets/sensors. The designed Artificial Intelligence (AI) enabled framework reads the smartphone sensors signal measurements to predict the grade of severity of the pneumonia as well as predicting the result of the disease.
I. INTRODUCTION
COVID-19 created urgent diagnostic pressure because existing approaches can be costly, time-consuming, imperfect, and difficult to scale. The paper proposes using smartphone sensors and computation to support diagnosis and monitor pneumonia severity.
- COVID-19 diagnosis relies on techniques including NAT and CT, with CT described as useful for assessing lung inflammation severity.
- Hospital CT demand can overload medical systems, expose patients to cross-infection risk, and contribute to delayed detection and treatment.
- RT-PCR and CT are not perfect, while medical detection kits are costly and require installation for diagnosis.
- Smartphones can capture, store, and analyze sensor and visual data, including serial CT images for monitoring lung inflammation.
- The proposed framework reads built-in smartphone sensors and scans CT images to identify viral pneumonia, predict disease results, and track severity.
II. BACKGROUND
The background reviews emerging AI and imaging approaches for COVID-19 diagnosis. It emphasizes that existing work and evidence were still limited because the disease was newly emerging.
- The literature on COVID-19 diagnosis was described as scarce because of the disease’s recent emergence.
- A reviewed deep-learning system detected COVID-19 from high-resolution CT images but relied only on CT data.
- Recent research indicated that COVID-19 detection results are more reliable when several methods are used jointly.
III. THE PROPOSED FRAMEWORK
The proposed framework combines smartphone sensors, symptom-processing algorithms, CT-image analysis, machine learning, and optional cloud exchange to diagnose COVID-19 and estimate disease severity.
- III. THE PROPOSED FRAMEWORK: The framework is intended as a low-cost, accessible solution for radiologists or smartphone users in emergency situations.
- III. THE PROPOSED FRAMEWORK: Algorithms estimate symptom levels and store each patient’s predicted symptom results as a record for machine-learning input.
- III. THE PROPOSED FRAMEWORK: The framework is organized into sensor reading, sensor configuration, symptom computation, and disease-prediction layers.
- III. THE PROPOSED FRAMEWORK: It collects CT images, camera video, accelerometer data during sit-to-stand, cough audio, and temperature measurements from smartphone sensors.
- III. THE PROPOSED FRAMEWORK: CT-image analysis compares lesion volume and density across scans, supporting assessment of pneumonia severity and lung inflammation.
- III. THE PROPOSED FRAMEWORK: Cloud exchange can expand the framework dataset and support transfer learning across smartphones and onboard sensors.
IV. TITLE
The proof-of-concept mobile app implements smartphone-based capture and submission of COVID-19 symptom data. Its functions cover fatigue, temperature, lung-image, breathing, and cough measurements.
- IV. TITLE: The implemented app captures COVID-19 symptom data and submits mobile sensory data to a cloud platform.
- IV. TITLE: Fatigue measurement records x, y, and z accelerometer axes, including background time-series capture for pattern construction.
- IV. TITLE: The app estimates body temperature using the rear camera and an index finger placed over the back flash.
- IV. TITLE: The app captures and stores lung X-ray or CT images for cloud-based segmentation and analysis, with image refinement options.
- IV. TITLE: Cough samples are recorded three times through the smartphone microphone and stored as a voice-signal file.
V. CONCLUSION
The study proposes a smartphone-sensor framework intended to address limitations of existing COVID-19 detection approaches. It combines multiple sensor readings through layered processing to predict disease and pneumonia severity.
- The framework uses smartphone sensor measurements as a low-cost COVID-19 diagnostic solution accessible to doctors, radiologists, and ordinary users.It is designed to operate on different smartphone platforms without external or additional sensors.
- Four layers read sensor measurements, configure sensors, compute disease symptoms, and predict disease using combined approaches.The final machine-learning stage could be further improved with cloud-based transfer learning.
- The framework is described as more reliable than state-of-the-art approaches because it combines readings from multiple sensors related to disease symptoms.