Review article

Digital Health
Volume 8: 1–10
Telemedicine and artificial intelligence to © The Author(s) 2022
Article reuse guidelines:
support self-isolation of COVID-19 patients: sagepub.com/journals-permissions
DOI: 10.1177/20552076221100634
Recent updates and challenges journals.sagepub.com/home/dhj

Jessica A Huang1, Intan R Hartanti1 , Michelle N Colin1

and Dian AE Pitaloka1,2

Abstract

Background: Asymptomatic and high-risk COVID-19 patients are advised to self-isolate at home. However, patients may not
realize that the condition is deteriorating until too late.
Objective: This study aims to review various artificial intelligence-based telemedicine research during the COVID-19 outbreak
and proposes a framework for developing telemedicine powered by artificial intelligence to monitor progression in COVID-19
patients during isolation at home. It also aims to map challenges using artificial intelligence-based telemedicine in the
community.
Methods: A systematic review was performed for the related articles published in 2019–2021 and conducted in the PubMed
and ScienceDirect database using the keywords “telemedicine,” “artificial intelligence,” and “COVID-19”. The inclusion cri-
teria were full-text articles and original research written in the English language.
Results: Thirteen articles were included in this review to describe the current application of artificial intelligence-based tele-
medicine during the COVID-19 pandemic. Various current applications have been implemented, such as for early diagnosis
and tracing of contact for the users, to monitor symptoms and decision-making treatment, clinical management, and virtual
and remote treatment. We also proposed the framework of telemedicine powered by artificial intelligence for support the
self-isolation of COVID-19 patients based on the recent update in technology. However, we identified some challenges for
using digital health technologies because of the ethical and practical use, the policy and regulation, and device use both
for healthcare workers and patients.
Conclusion: Artificial intelligence promises to improve the practice of medicine in various ways. However, practical applica-
tions still need to be explored, and medical professionals also need to adapt to these advances for better healthcare delivery
to the public.

Keywords

Artificial intelligence, telemedicine, COVID-19, self-isolation patient

Submission date: 28 March 2022; Acceptance date: 27 April 2022

1
Department of Pharmacology and Clinical Pharmacy, Faculty of Pharmacy,
Introduction Universitas Padjadjaran, Sumedang, Indonesia
2
Center of Excellence in Higher Education for Pharmaceutical Care
It has been highlighted that the emergence of the corona-
Innovation, Universitas Padjadjaran, Sumedang, Indonesia
virus disease 2019 (COVID-19) pandemic in the world,
Corresponding author:
including in Indonesia, posed a new challenge to the Dian AE Pitaloka, Department of Pharmacology and Clinical Pharmacy,
efforts to control infectious diseases. COVID-19 is charac- Faculty of Pharmacy, Universitas Padjadjaran, Sumedang 45363, Indonesia.
terized by a range of clinical signs of asymptomatic Email: [email protected]

Creative Commons NonCommercial-NoDerivs CC BY-NC-ND: This article is distributed under the terms of the Creative Commons Attribution-
NonCommercial-NoDerivs 4.0 License (https://creativecommons.org/licenses/by-nc-nd/4.0/) which permits non-commercial use, reproduction
and distribution of the work as published without adaptation or alteration, without further permission provided the original work is attributed as specified on
the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage).
2 DIGITAL HEALTH

infection.1 COVID-19 can also be particularly dangerous 2. Identify challenges and recommend solutions for
for certain high-risk patients, including elderly patients researchers to develop an advanced AI-based telemedi-
and those with specific pre-existing health conditions.2 In cine system for quarantining of COVID-19 patients
practice, asymptomatic and high-risk patients self-isolate based on the integration of recent studies.
at home and are expected to call their doctor if their condi- 3. Helps policymakers and clinicians identify the key
tion worsens.3,4 However, patients can begin to deteriorate points for implementing AI-based telemedicine in
without notice, and by the time they realize that something patients.
is mistaken, the chance for early life-saving treatment may
have been missed.4 Based on the issue, it is necessary to
develop methods to communicate efficiently with asymp-
tomatic and high-risk COVID-19 patients under self- Methods
isolation at home. To give a brief description of the current application of
Telemedicine is a term that was invented back in the AI-based telemedicine during the COVID-19 pandemic in
1970s, which means “healing from afar” and refers to infor- the medicinal field, we conducted a systematic review in
mation communication technology used to improve accordance with the Preferred Reporting Item for
society’s health and well-being by enhancing access to Systematic reviews and Meta-Analyses.15 It includes
healthcare information.5 It allows clinicians to provide various processes: information source and eligibility cri-
health services when social distancing is needed without teria, study selection, and data extraction.
direct contact with patients.6 The most commonly used
technologies are voice and video calls.7 Telemedicine
allows using technology without requiring healthcare pro-
fessionals and patients to be in the same place.8 Information source and eligibility criteria
Telemedicine combines convenience, comfortability, and The selection for the related article published in 2019–2021
low-cost access to information regarding health and com- was conducted in the PubMed and ScienceDirect database
munication technology to support health practice remotely using the keywords “telemedicine,” “artificial intelligence,”
through phone, video, email, and mobile health applica- and “COVID-19.” The inclusion criteria were full-text arti-
tions.8,9 However, data security problems, poor user educa- cles written in the English language and original research
tion background, and lack of telemedicine features have studies. We excluded the articles that did not describe the
been found as the problems faced today in the application use of AI-driven technology in telemedicine in
of telemedicine in the community.10 COVID-19 cases or other diseases during the COVID-19
Although artificial intelligence (AI) was initially intro- pandemic, protocol studies, conference abstracts, case
duced > 60 years ago, the rapid evolution of AI-based tech- reports, commentaries, editorials, and reviews.
nology and applications occurred after the improvement of
graphic processing units in the 2010s.11 AI is the term used
to describe computers and technology to simulate intelli-
gent behavior and critical thinking comparable to a Study selection
human being.12 John McCarthy first defined AI in 1956 The search process of this study consisted of three sub-
as the science and engineering of making intelligent processes, they are a collection of articles, scanning of
machines.13 Recently, AI algorithms have been studied title and abstract, and reading of the full text. Each article
and established in various medical fields to facilitate was assessed by two reviewers to ensure the inclusion cri-
exam interpretations, improve the accuracy of diagnoses, teria and the extracted data. Any disagreement between
and reduce time and human resource consumption.14 the two reviewers was resolved by a senior reviewer. The
This review focuses on recent advances in AI technology senior reviewer also evaluated the quality of the included
and its current application in telemedicine during the articles to ensure their significance to this review. A total
COVID-19 era. The proposed framework for the devel- of 626 publications were found, 13 of which met the inclu-
opment of telemedicine powered by AI to support self- sion criteria (Figure 1).
isolation in COVID-19 patients is provided in the follow-
ing description. It also identifies the challenges and
acceptable use of AI in telemedicine during the
COVID-19 era. The contributions of this article are sum- Data extraction
marized as follows: We extracted the required data and collected it for further
analysis manually using an Excel spreadsheet. The numer-
1. Provide a clear insight into the current implementation ous parameters such as title, year, type of articles, and the
of AI in telemedicine for various diseases during the result were selected for achieving the key points of the sys-
COVID-19 outbreak. tematic reviews.
Huang et al. 3

Figure 1. Flow diagram of inclusion and exclusion of studies. Reasons for exclusion are conference abstract, case reports, commentaries,
editorials, protocols, and reviews (narrative review, systematic review, and meta-analysis).

Telemedicine and AI during COVID-19 era: Current several uses: early diagnosis and tracing of contacts, mon-
application itoring symptoms and decision-making treatment, clinical
management, and virtual and remote treatment (Table 1).
AI is growing into the public health sector and will signifi-
cantly impact every aspect of primary care. AI in medicine
can be divided into two subtypes: virtual and physical. The
virtual component is represented by Machine or Deep Early diagnosis and tracing of contact for the users
Learning, represented by mathematical algorithms that Miyake et al.18 conducted a study of the active telemedicine
improve learning through experience. It enables “systems method and computed tomography (CT) protocol to reduce
thinking” about healthcare; it not only focuses on the clas- the risk of infection in a medical and non-medical team
sical interactions between patients and providers but also dealing with COVID-19 patients. It showed no positive
considers larger-scale organizations and cycles. The serum-specific antibody testing and polymerase chain reac-
virtual part ranges from electronic health record systems tion testing results for SARS-CoV-2 in a group of COVID
to neural network-based guidance in treatment decisions.13 doctors. Furthermore, the 36-item short-form of the
The second form of application of AI in medicine Medical Outcome Study Questionnaire showed no deterior-
includes physical objects, medical devices, and increasingly ation in physical and mental quality of life status. No
sophisticated robots in care delivery.16 The most promising in-hospital infection occurred during the study period.
approach is using robots as helpers; for example, a robot The combination strategy seems acceptable for both the
companion for the aging population with cognitive protection and stress relief among the medical staff.
decline or limited mobility. Robots are used in surgery as Alodat et al.19 created real-time telemedicine powered
assistant surgeons or even as solo performers.17 The phys- by various deep learning-based real-time telemedicine for
ical part deals with robots assisting in performing surgeries, adapting and managing the COVID-19 pandemic crisis. It
intelligent prostheses for disabled people, and elderly care. was reported that Convolutional Neural Networks using
The current research focuses on utilizing AI develop- Tensorflow (CNN-TF) model were capable of discriminat-
ments to support various diseases, including COVID-19 ing between those with positive cases of COVID-19 and
during the Coronavirus outbreak. It is categorized under those with negative cases. Another model, the Random
 4
Table 1. Implementations of recent technologies in medicine during the COVID-19 era.

References Subject Study design Highlight Function

Miyake et al.18 Ten clinicians from the COVID team who Participants were evaluated the serum-specific Combination of telemedicine and CT protocol Contact tracing
used a CT-first triage protocol and antibodies for SARS-CoV-2 at the initial and end was found to help the medical staff protect
telemedicine for 165 individuals of the study, PCR at the end of the study, and from the risk of infection who dealt with
inpatients and outpatients 36-item short-form of the Medical Outcome Study COVID-19 patients
Questionnaire

Alodat19 Eleven participants were randomly Eligible participants have made chest X-ray images Deep learning-based real-time medicine Diagnosis
chosen from different areas who had to analyze and predict COVID-19. Four robust using Convolutional Neural Networks using
respiratory symptoms models in real-time telemedicine were Tensorflow (CNN-TF) model was able to
developed and used to assist those X-ray images discriminate between positive and negative
COVID-19 cases

Bassam et al.20 The study did not directly simulate The proposed system is implemented with IoT-based wearable monitoring device was Monitoring signs
humans. The AI model was trained three-layered functionalities as wearable IoT designed to measure various vital signs and symptoms
using a recording signal for both sensor layer, cloud layer with API and Android related to COVID-19, such as body
cough and noise for 10 s web layer for mobile phones. These integrated temperature, oxygen saturation in the
systems were trained to predict signs of COVID-19 blood, heartbeat monitoring, respiratory
patients system

Sharma et al.21 The study did not directly simulate Proposed diagnostic model was based on Bio wearable sensor system based on ontology Decision-making
humans. Simulations were done using electrocardiogram, photoplethysmography, method used sensory 1D biomedical signals treatment
the R software and cooza simulator temperature, and accelerometer and was to monitor remote patients and provide
validated using 10-fold cross-validation. medical help to distant locations
Accuracy of the model was trained to classify the
patients into an infected and non-infected

Faris et al.22 A total of 246,814 consultations and 1206 The system was built from a fusion of ML models Combination of telemedicine and Decision-making
diagnoses were collected from Altibbi trained based on two modalities: the symptoms computer-aided intelligence was reported treatment
company and the medical questions of the patients to help doctors and clinicians in making
correct treatment decisions based on the
given symptoms and patients’ questions

Dawoodbohy 9 mental health healthcare practitioners Combination of a narrative literature review and AI integrated with telemedicine is utilized to Preventive and
et al.23 (HCPs) and 11 artificial intelligence pilot interview was conducted with AI and mental improve the delivery of preventive and clinical
(AI) experts health experts personalized care for mental health management
condition patients

(continued)
DIGITAL HEALTH
 Huang et al.

Table 1. Continued.
References Subject Study design Highlight Function

Miyake et al.18 Ten doctors of the COVID team, who used Evaluation of serum-specific antibodies for Deep learning-based real-time medicine Clinical
a CT-first triage protocol and SARS-CoV-2 at the initial and end of the study, using the Random Forest classification management
telemedicine for 165 individuals PCR result at the end of the study, and 36-item model could predict patients with a risk of
inpatients and outpatients short-form of the Medical Outcome Study death and provide appropriate healthcare
Questionnaire

Adly et al.24 In total, 60 participants with stage 1 Group A received oxygen therapy with bilevel A newly developed Telemanagement with Clinical
pneumonia caused by SARS-CoV-2 positive airway pressure ventilation, and group B home-based oxygen therapy with bilevel management
infection received osteopathic manipulative respiratory positive airway pressure was found to be
and physical therapy techniques. Arterial blood more effective for prophylactic treatment in
gases, pH, vital signs, and chest CT scans were early-stage COVID-19 pneumonia
used for follow-up and assessment

Keenan et al.25 Four individuals (mean age, 73.8 years) Eligible participants performed daily self-imaging Telemedicine powered by NVHO-based deep Virtual and remote
with neovascular age-related macular with the NVHO coherence tomography for 1 learning is used to evaluate neovascular treatment
degeneration (one or both eyes) month. The macular cube scans were uploaded age-related macular degeneration patients
undergoing anti-vascular endothelial automatically to the Notal Health Cloud. They who do a home treatment. It may allow
growth factor therapy underwent evaluation separately by the Notal highly personalized retreatment decisions,
OCT Analyzer and human expert graders for fluid with fewer unnecessary injections and clinic
presence, segmentation, and volume visits

Chae et al.26 In total, two groups of stroke survivors An HBR system involves an off-the-shelf A smartphone app equipped with an ML Virtual and remote
contain 17 and 6 participants were smartwatch, a smartphone, and created with accelerometer and gyroscope treatment
enrolled for statistical analysis custom-developed apps. A convolutional neural data performed effectively and improved
network was used to train the ML algorithm for the wolf motor function in HBR chronic
detecting home exercises stroke patients

Note: CT: computed tomography; API: application peripheral interface; NVHO coherence tomography: Notal Vision Home Optical coherence tomography; HBR: home-based rehabilitation; ML: machine learning.
5
6 DIGITAL HEALTH

Forest classification, was found to predict COVID-19 involvement from the patient, such as completing surveys
patients at risk of death and provided appropriate healthcare or contributing audio samples using digital apps. They are
assistance.19 Chest X-ray image analysis was also per- then analyzed with AI to tailor services to the individual
formed to provide a differential diagnosis of respiratory dis- by further engaging patients in their recovery process.
eases related to COVID-19. However, “passive data” can also be collected from
digital devices via mobile applications. The smartphone
application will collect data from mobile sensors and
Monitor symptoms and decision-making treatment phone usage, such as location and SMS logs, to monitor
Bassam et al.20 designed the Internet of Things (IoT) based the state of a patient’s disease.23
wearable monitoring device to measure various vital signs Adly et al.24 compared the nonpharmacological respira-
related to COVID-19. The system will automatically alert tory treatment methods for home-isolated COVID-19
the concerned clinical authorities about quarantine viola- patients using a new Telemanagement healthcare system.
tions for potentially infected patients by monitoring their It was found that home-based oxygen therapy with bilevel
real-time GPS data. The proposed approach is implemented positive airway pressure can be a more effective prophylac-
with three-layered functionalities: wearable IoT sensor tic treatment than osteopathic manipulative respiratory and
layer, cloud layer with Application Peripheral Interface, physical therapy techniques because it can impede exacer-
and android web layer for mobile phones. This design bation of early-stage COVID-19 pneumonia.
may serve as an essential part of COVID-19 patients and
can be used as a monitoring device to track the health and
recovery of a COVID-19 patient. Virtual and remote treatment
Another IoT-based remote access was also developed by
The system’s performance of Notal Vision Home Optical
Sharma et al.21 They combined it with an alarm-enabled bio (NVHO) coherence tomography, telemedicine infrastruc-
wearable sensor system for the early detection of ture for automated data upload, and deep learning algorithm
COVID-19. The design was based on the ontology
for automated optical coherence tomography of self-
method using sensory 1D biomedical signals such as elec- imaging at home were evaluated in longitudinally neovas-
trocardiogram, photoplethysmography, temperature, and cular age-related macular degeneration patients. It was
accelerometer to monitor remote patients and provide
found that 87.9% of patients initiated self-imaging and
medical help to distant locations. It was reported that the 97.6% of these had satisfactory image quality. Thus, it
efficiency of the proposed model was 96.33% to distinguish may allow highly personalized retreatment decisions with
between COVID-19 patients and suspected COVID-19
fewer clinic visits and alternative disease monitoring
patients. during the COVID-19 pandemic.25
A recent study proposes a computer-aided intelligent A previous study developed a home-based rehabilitation
diagnosis decision support system as part of telemedicine
(HBR) system that could recognize and record the type and
to help clinicians make a correct decision. The design frequency of rehabilitation exercises conducted by the user
encompasses a fusion of machine learning (ML) models using a smartwatch and smartphone app equipped with an
trained based on two modalities: the patients’ symptoms
ML algorithm in chronic stroke survivors during the
and the medical questions. The output of the combination COVID-19 pandemic. It was found that the ML model
showed promising predictive ability with a classification created with accelerometer and gyroscope data was the
accuracy of 84.9%, indicating the potential of the model
most accurate (99.80%). It also performed a significant
in predicting the diagnosis of possible patient conditions improvement (17/22) in the Wolf Motor Function Test for
based on the given symptoms and patients’ questions that HBR patients.26
consequently could aid clinicians in making the right
decisions.22
Designing telemedicine powered by AI to support
Clinical management self-isolation of COVID-19 patients
Semi-structured interviews were conducted with AI and A built-in smartphone sensor can be implemented to test the
mental health experts. The data were then used to analyze capturing functions of COVID-19 symptoms in the diag-
and synthesize gathered data to construct an enhanced nosing process. The parameters input are CT scan images
model with potential AI solutions integrated with telemedi- of the lung, 30-s sit-to-stand, cough voice samples, and fin-
cine. The result showed that AI tools could enable gerprint on the screen. The smartphone’s camera, micro-
improved delivery of preventive and personalized care by phone, and temperature fingerprint sensor can detect all
opening up new data collection and analysis avenues to these parameters to calculate abnormal sub-image lung
enhance understanding of mental health conditions. For shapes, nausea or fatigue level, cough level, and fever
example, “active data” will be collected with active level. Based on a developed algorithm, symptom prediction
Huang et al. 7

Figure 2. The framework of a smartphone application for supporting self-isolation COVID-19 patients.

will be analyzed by ML techniques for positive or negative biosensor will be integrated with a specially designed
COVID-19 initial results.27 Inception Residual Recurrent smartphone application, and the patient can report their
Convolutional Neural Network with Transfer Learning symptoms and record their cough sound using the smart-
(TL) can also be integrated into telemedicine to evaluate phone. The analytics engine will process these multidimen-
X-ray and CT scan images to detect COVID-19 cases sional physiology parameters to detect subtle physiological
(Figure 2).28 changes preceding critical events, thereby enabling clini-
A deep-learning algorithm can be used to diagnose and cians to review and intervene promptly.31
stratify the risk of COVID-19 from lung sounds. It is
based on the fact that abnormal lung sounds have character-
istic frequencies and duration, differentiating them from
normal lung sounds.29 Several data augmentation techni- Challenges and acceptable use of telemedicine
ques will be explored, such as amplitude scaling, pitch and AI
shift, and random time shift. This data set will then be fed The COVID-19 pandemic has become an additional accel-
into a convolutional neural network with max-pooling erator of changes in terms of the healthcare system digitiza-
and dropout before binary classification by a support tion. It brings healthcare workers closer to the patient,
vector machine into positive versus negative COVID-19 supporting the exchange of medical data between the
test results (diagnostic model) or hospitalization versus doctor and the patient and improving the circulation of
outpatient.30 information. However, digital healthcare such as telemedi-
Wearable biosensors can be developed to continuously cine has been debatable because of privacy and security,
monitor multidimensional physiological parameters for data format and management, synchronization, over-
the early detection of COVID-19 clinical progression. The tracking, and lack of proper development and implementa-
physiological parameters monitored include skin tempera- tion guidelines that impact their efficacy and adoption.32
ture, respiratory rate, blood pressure, pulse rate, blood This is exacerbated by hasty, poor user engagement, ill-
oxygen saturation, and daily activities rather than relying prepared, or poorly communicated implementation of tele-
solely on temperature measurement. The wearable medicine that ultimately influences public trust33
8 DIGITAL HEALTH

Trust in the age of digital healthcare needed. Device accuracy and improvement are also barriers
identified by patients who want to use a digital device.
It is confirmed that the challenges of digital technology
Another study shows that many digital health technologies
systems in the healthcare sector are not only technical
frequently did not provide a patient with the choice of a
since they are rooted and linked to the need to create a
clinician or transparent clinician credentials and offered
high level of trust in such scenarios. It is the foundation
diagnoses without adequate medical history-taking.39
for the ethical and practical use of new digital tools in
increased awareness and the more active role of patients
in the treatment process. Even though a number of studies Limitation
concerning trust in technology have been undertaken so
However, our study has limitations. Although the selected
far, the problem of trust in such systems is still up-to-date
articles in our study supported the study approach, we only
and requires further exploration.34
focused on two databases, PubMed and ScienceDirect, to
Mbunge et al.35 identified transparency, regular engage-
identify the articles that met our article selection criteria.
ment, and possible user parameters to boost public trust
Thus, other related articles may not have been identified
However, poor communication and hasty implementation
because they are not in those two databases. Despite the limi-
of digital healthcare will undermine public confidence and
tation, our study has described the current implementation of
further risk efficacy. Despite the beneficial impact of
AI in telemedicine in various diseases during the COVID-19
digital healthcare, its integration into clinical practice
outbreak. It also gives clear insights into the policymakers
should be carefully observed to avoid the potential violation
and clinicians to identify the key point for implementing
of ethical practices and regulatory policies.
AI-based telemedicine in patients.

The policy and regulation of digital healthcare Conclusion

The future of digital healthcare will be significantly influ- AI promises to change the practice of medicine so far in
enced by a complex set of policy, legal, and regulatory fra- various ways. However, many practical applications are
meworks. An understanding of these frameworks will help still in their infancy and need to be explored and developed
clinicians more effectively adopt and implement telemedi- better. Medical professionals also need to understand and
cine platforms. Chen et al.36 analyzed the impact of adapt to these advances for better healthcare delivery to
policy instruments to promote telemedicine interventions the public. Moreover, support from a set of policies, regu-
for COVID-19. It was reported that around 62% of policies latory frameworks, and data transparency are important
aimed to strengthen clinical service, and more telemedicine keys leading to the more accessible use of telemedicine.
technology interventions were implemented in clinical ser-
vices (38.7%). However, emergency planning and pre-
Acknowledgment: This study was supported by the Center of
paredness had the least self-reported telemedicine Excellence in Higher Education for Pharmaceutical Care
intervention (27.1%). The adoption of telemedicine has Innovation, Universitas Padjadjaran.
been limited by licensure requirements, which impose com-
pliance burdens on clinicians. Policymakers could reduce
this burden by providing a telemedicine practitioner licens- Conflict of interest: The authors declared no potential conflicts of
interest with respect to the research, authorship, and/or publication
ing pathway that would enable clinicians to practice tele-
of this article.
medicine more easily.37

Contributorship: All authors make authorship contributions

Challenges in device use including conceptualization, expertise, technicalities, preparation
of the initial draft, review, and revision of the final version of
Many digital health technologies for various diseases are this study.
currently available. However, adaptation into clinical prac-
tice by healthcare workers as well as patients is unclear. A
recent electronic survey found that patients are likely to use Ethical approval: Not applicable; no human subjects
involvement.
digital healthcare recommended by healthcare practitioners
(HCPs), as reported by Ding et al.38 for digital health
devices for atrial fibrillation (AF). A total of 53.4% of Funding: The authors received no financial support for the
patients wanted guidelines from their HCP for optimal research, authorship, and/or publication of this article.
use. Around 64% of patients wanted more data comparing
the accuracy of digital devices to conventional devices for ORCID iDs: Intan R Hartanti https://orcid.org/0000-0003-
AF monitoring. It can be observed that expanding familiar- 0538-5352
ity with digital technology to general practitioners is Dian AE Pitaloka https://orcid.org/0000-0002-8187-8864
Huang et al. 9

References 20. Assam NA, Hussain SA, Qaraghuli AA, et al. IoT based wear-
able device to monitor the signs of quarantined remote
1. Kronbichler A, Kresse D, Yoon S, et al. Asymptomatic
patients of COVID-19. Inform Med Unlocked 2021; 24:
patients as a source of COVID-19 infections: a systematic
100588.
review and meta-analysis. Int J Infect Dis 2020; 98: 180–186.
21. Sharma N, Mangla M, Mohanty SN, et al. A smart ontology-
2. Guan WJ, Ni ZY, Hu Y, et al. Clinical characteristics of cor-
based IoT framework for remote patient monitoring. Biomed
onavirus disease 2019 in China. N Engl J Med 2020; 382:
Signal Process Control 2021; 68: 102717.
1708–1720.
22. Faris H, Habib M, Faris M, et al. An intelligent multimodal
3. CDC. Healthcare workers. Centers for Disease Control and
medical diagnosis system based on patients’ medical ques-
Prevention, https://www.cdc.gov/coronavirus/2019-ncov/
tions and structured symptoms for telemedicine. Inform Med
hcp/clinical-guidance-management patients.html (2020,
Unlocked 2021; 23: 100513.
accessed 30 October 2021).
23. Dawoodbhoy FM, Delaney J, Cecula P, et al. AI in patient
4. You Y, Yang X, Hung D, et al. Asymptomatic COVID-19
flow: applications of artificial intelligence to improve patient
infection: diagnosis, transmission, population characteristics.
flow in NHS acute mental health inpatient units. Heliyon
BMJ Support Palliat Care 2021; 0: 002813.
2021; 7: e06993.
5. WHO. Telemedicine: Opportunities and Developments in
24. Adly AS, Adly MS and Adly AS. Telemanagement of home-
Member State | Regional Office for Africa, https://www.
isolated COVID-19 patients using oxygen therapy with non-
afro.who.int/publications/telemedicine-opportunities-and-
invasive positive pressure ventilation and physical therapy
developments-member-state (2009, accessed 6 November
techniques: randomized clinical trial. J Med Internet Res
2021).
2021; 23: e23446.
6. CDC. Telemedicine, https://www.cdc.gov/coronavirus/2019-
25. Keenan TDL, Goldstein M, Goldenberg D, et al. Prospective,
ncov/global-covid-19/telemedicine.html (2020, accessed 30
longitudinal pilot study: daily self-imaging with patient-
October 2021).
operated home OCT in neovascular age-related macular
7. Tozour JN, Bandremer S, Patberg E, et al. Application of tele-
degeneration. Ophthalmol Sci 2021; 1: 100034.
medicine video visits in a maternal-fetal medicine practice at
26. Chae SH, Kim Y, Lee KS, et al. Development and clinical
the epicenter of the COVID-19 pandemic. Am J Obstet
evaluation of a web-based upper limb home rehabilitation
Gynecol MFM 2021; 3: 100469.
system using a smartwatch and machine learning model for
8. Banks J, Corrigan D, Grogan R, et al. LoVE in a time of
chronic stroke survivors: prospective comparative study.
COVID: clinician and patient experience using telemedicine
JMIR MHealth UHealth 2020; 8: e17216.
for chronic epilepsy management. Epilepsy Behav EB 2021;
27. Maghdid HS, Ghafoor KZ, Sadiq AS, et al. A novel
115: 107675.
AI-enabled framework to diagnose coronavirus COVID 19
9. Vidal-Alaball J, Acosta-Roja R, Pastor Hernández N, et al.
using smartphone embedded sensors: design study.
Telemedicine in the face of the COVID-19 pandemic. Aten
ArXiv200307434 Cs Q-Bio 2020; 0: 180–187.
Primaria 2020; 52: 418–422.
28. Alom MZ, Rahman MMS, Nasrin MS, et al. COVID_MTNet:
10. Medical Clinic Express. 24 Hr TeleHealth Care, https://
COVID-19 detection with multi-task deep learning
livetelemdcare.com (2020, accessed 30 October 2021).
approaches. ArXiv200403747 Cs Eess 2020; 3: 1–10.
11. LeCun Y, Bengio Y and Hinton G. Deep learning. Nature
29. Bohadana A, Izbicki G and Kraman SS. Fundamentals of lung
2015; 521: 436–444.
auscultation. N Engl J Med 2014; 370: 744–751.
12. Amisha MP, Pathania M, et al. Overview of artificial intelligence
30. Glangetas A, Hartley MA, Cantais A, et al. Deep learning
in medicine. J Fam Med Prim Care 2019; 8: 2328–2331.
13. Mintz Y and Brodie R. Introduction to artificial intelligence in diagnostic and risk-stratification pattern detection for
medicine. Minim Invasive Ther Allied Technol 2019; 28: 73–81. COVID-19 in digital lung auscultations: clinical protocol for
14. Jheng YC, Kao CL, Yarmishyn AA, et al. The era of artificial a case-control and prospective cohort study. BMC Pulm
intelligence-based individualized telemedicine is coming. J Med 2021; 21: 03.
Chin Med Assoc 2020; 83: 981–983. 31. Wong CK, Ho DTY, Tam AR, et al. Artificial intelligence
15. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA mobile health platform for early detection of COVID-19 in
2020 statement: an updated guideline for reporting systematic quarantine subjects using a wearable biosensor: protocol for
reviews. Br Med J 2021; 372: 71. a randomised controlled trial. BMJ Open 2020; 10: e038555.
16. Cornet G. Chapter 4. Robot companions and ethics: a pragmatic 32. Min-Allah N, Alahmed BA, Albreek EM, et al. A survey of
approach of ethical design. J Int Bioethique 2013; 24: 49–58. COVID-19 contact-tracing apps. Comput Biol Med 2021;
17. Larson JA, Johnson MH and Bhayani SB. Application of surgi- 137: 104787.
cal safety standards to robotic surgery: five principles of ethics 33. Ranisch R, Nijsingh N, Ballantyne A, et al. Digital contact
for nonmaleficence. J Am Coll Surg 2014; 218: 290–293. tracing and exposure notification: ethical guidance for trust-
18. Miyake S, Higurashi T, Kato H, et al. Evaluation of a combin- worthy pandemic management. Ethics Inf Technol 2020; 23:
ation protocol of CT-first triage and active telemedicine 1–10.
methods by a selected team tackling COVID-19: an experi- 34. Wiś niewska J and Róż ycka M. The problem of trust in
mental research study. J Infect Public Health 2021; 14: innovative ICT technologies used in e-health systems. The
1212–1217. case study of private health care units located in Szczecin.
19. Alodat M. Using deep learning model for adapting and man- Procedia Comput Sci 2021; 192: 3647–3656.
aging COVID-19 pandemic crisis. Procedia Comput Sci 35. Mbunge E, Millham RC, Sibiya MN, et al. Framework for
2021; 184: 558–564. ethical and acceptable use of social distancing tools and
10 DIGITAL HEALTH

smart devices during COVID-19 pandemic in Zimbabwe. 38. Ding EY, Svennberg E, Wurster C, et al. Survey of current
Sustain Oper Comput 2021; 2: 190–199. perspectives on consumer-available digital health devices
36. Chen M, Xu S, Husain L, et al. Digital health interventions for for detecting atrial fibrillation. Cardiovasc Digit Health J
COVID-19 in China: a retrospective analysis. Intell Med 2020; 1: 21–29.
2021; 1: 29–36. 39. Resneck JS, Abrouk M, Steuer M, et al. Choice, transparency,
37. Shachar C, Engel J and Elwyn G. Implications for telehealth coordination, and quality among direct-to-consumer telemedi-
in a postpandemic future: regulatory and privacy issues. cine websites and apps treating skin disease. JAMA Dermatol
JAMA 2020; 323: 2375–2376. 2016; 152: 768–775.