COVID-19 lessons learned: medical devices at the core of global healthcare. A foreword on new challenges for expert review of medical devices!

KEYWORDS:

• digital health
• artificial intelligence
• medical devices
• oncology
• telemedicine
• telehealth
• COVID-19

1. Introduction

One of the most intriguing aspects in the famous novel ‘The name of the Rose’ is the final Latin hexameter inspiring the book’s title: ‘stat rosa pristina nomine, nomina nuda tenemus.’ Of the various interpretations given to this suggestive sentence, the most plausible is that all the departed things leave pure names behind them. Names are important and we should stop and think about names’ origin and meanings. Since I have been humbled to be appointed as the Editor-in-Chief of Expert Review of Medical Devices, I find myself thinking about the origin and meaning of the term ‘device.’ According to the site www.etymonline.com, the modern term ‘device’ derives from the old French devis ‘division, separation; disposition’ that, in its turn, originates from the Latin verb ‘dividere’ meaning ‘to divide or arrange in small parts.’ When looking back to the long pathway leading from the ancient ‘dividere’ to the modern ‘device,’ one might be not so surprised, considering all the developments that have occurred in medical devices over the past several decades, especially through the implementation of sophisticated, flexible, adaptable, and sometimes even invisible, technologies [1–3].

Technological advances in medicine have been gaining an ever-increasing importance in leaving the COVID-19 pandemic behind [4]. As a part of the global response to fight pandemic, we have witnessed an incredible enhancement of interest in medical devices: 9954 results can be retrieved when typing on PubMed search engine the combined terms ‘medical device’ AND ‘COVID-19’ (last accessed on 30 November 2022). In the following foreword, we briefly cover two paradigmatic fields of study and research that have emerged as particularly relevant during the pandemic: predictive models based on artificial intelligence (AI) and so-called ‘digital health.’

2. Artificial intelligence for medical imaging

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of COVID-19, can attack pulmonary alveolar cells binding to angiotensin-converting enzyme 2 (ACE2) receptors, a critical step for the virus to enter into target cells and cause infection [5]. Inflammatory responses to SARS-CoV-2 infection can result in a severe form of pneumonia, the most relevant cause of death during the pandemic. Many efforts have been made to diagnose COVID-19 lung disease at an early stage, through chest X-ray (CXR) or computed tomography (CT).

AI-based techniques have been applied to aid the diagnosis of COVID-19 pneumonia by CXR, the 1^st line and cheapest imaging modality, available worldwide. In this regard, a CXR-based deep neural Network (CXR-Net) has been recently developed, on the basis of an ‘Encoder-Decoder-Encoder’ architecture. When tested on real world CXR datasets obtained from public and private sources, the CXR-Net algorithm proved it was capable at identifying COVID-19 pneumonia with a sensitivity, specificity, and accuracy of 0.992, 0.998, 0.985, respectively [6]. Aside from diagnosis, imaging has been also applied for COVID-19 patients’ prognostication, particularly for the prediction of disease progression [6]. It has been reported, in fact, that the quantification of the extent of lung involvement in COVID-19 pneumonia by chest CT examination (CT score) strictly correlates with laboratory and clinical parameters with meaningful impact on clinical decision [7,8]. However, manual semi-quantitative lung scoring on CT images is a time-consuming approach, requiring well-trained dedicated radiologists. In a recently published multi-institutional study, three different methods for CT scoring were compared: 1) semi-quantitative manual scoring by three experienced radiologists; 2) deep-learning-based segmentation of ground-glass and consolidation areas obtained by CT Pulmo Auto Results prototype plugin on IntelliSpace Discovery (Philips Healthcare, The Netherlands); 3) threshold-based segmentation of involved lung utilizing an open-source tool [9]. Manual segmentation showed meaningful limitations, especially in case of more extent disease; by contrast, both deep learning-based and threshold segmentation had excellent performance for an accurate CT scoring, abolishing inter-readers’ variability and significantly impacting on time-saving of trained medical personnel.

3. Digital health: no one left behind

Social distancing has represented a necessary measure to help contain and arrest COVID-19 spreading [10]. Nevertheless, although widely justified, this extreme measure has entailed a relevant hardship to detect new-onset oncological and non-oncological diseases, as well as monitoring patients affected by chronic pathologies. This emergency led to an outburst in the field of digital health technologies, particularly concerning telemedicine and remote/virtual care, in order to support and sometimes replace traditional healthcare [11]. A first crucial point is the definition itself of ‘digital health.’After having reviewed 1527 records, Fatehi and coworkers found that the dominant concept in digital health is represented by mobile health (mHealth), that in its turn encompasses other issues such as telehealth, e-Health, and the already mentioned AI-based applications to healthcare [12].

Telehealth, which can be considered as a cornerstone of digital health, includes all the healthcare facilities carried out through an exchange of health information using mobile smart connected devices and is articulated in a huge range of activities such as tele-expertise, teleconsultation, telecare etc … In a recently published review on this topic from Bouabida and coworkers, three different levels of telehealth were identified on the basis of the employed technologies: 1) technologies that allow communication but don’t allow patients’ self-measurements; 2) technologies that allow, aside exchange of health-related data, simple self-measurements (i.e. body temperature, blood pressure); 3) sophisticated technologies aimed to perform remote monitoring and processing of clinical data [13].

Digital health has been applied in several oncological and non-oncological settings. Schinasi and colleagues, for example, carried out an interesting cross-sectional survey of clinicians after implementing telehealth practice at an independent children’s hospital [14]. Aside from assessing confidence and concerns about the use of digital health in pediatrics during pandemic, the authors also checked clinicians’ intention to eventually continue the practice beyond COVID-19. The main drawbacks in telehealth practice registered in the survey were reliability of internet (82%) and the limitations to assess patients’ physical condition by video (73.8%). Regarding the intention to continue telehealth activities, the majority of survey-responders were willing to provide patients’ care by telehealth as long as payers continued to reimburse digital visits. The authors found that providers who performed a higher number of video-visits also reported greater ease of incorporating telehealth into routine practice, therefore suggesting that dedicated educational frameworks are needed to support the widespread of digital health.

4. Conclusions and future outlook

Looking beyond its dramatic impact on everyone’s life, COVID-19 taught some relevant lessons to the medical community. First of all, medical technologies and devices were essential to support healthcare in emergency conditions. AI and its various applications have the potential to be successfully applied for the management of a huge quantity of data, such as those deriving from diagnostic imaging, with great accuracy and a meaningful saving of time. This primary lesson should not remain unheard since it holds the promise to move the field forward. As an example, in elderly populations AI-based techniques have been employed to reach a more precise classification of subjects with cognitive deterioration, on the basis of imaging and laboratory findings [15,16]. In addition, another AI-based discipline, namely radiomics, has been arousing great interest concerning its capability to extract from images quantitative data, undetectable to the naked eye and potentially useful for aiding diagnosis and assessing response to therapy [17]. However, the majority of AI-based approaches are at an embryonic stage as they still need a large-scale validation, as well as standardized data collection, evaluation, and reporting criteria to become ‘a mature’ * [18] discipline and find a well-established collocation in clinical practice [19].

On the other side, digital health is not without costs. It has to be highlighted, in fact, that during the pandemic socioeconomic factors and rurality had a significant impact on determining access to telehealth services. In a retrospective study on digital health access inequities during COVID-19, some factors such as old age, self-pay or other insurance vs. commercial insurance, Black or Asian vs. White ethnicity and non-English preferred languages resulted associated with a lower probability to have a telehealth visit [20]. Of note, living in a zip code with lower internet access was correlated with a reduced access to digital health too. Furthermore, as previously mentioned, specific educational pathways would be necessary to support and incentivize the use of telehealth in different socio-economical settings.

COVID-19 impressively boosted innovation in the field of medical devices. In this regard, many relevant applications should be mentioned, such as the measurement systems to assess the distribution of aerosols and droplets in indoor environments, the sophisticated biotechnologies employed for SARS-CoV-2 genome sequencing, or the veno-venous extracorporeal membrane oxygenation (VV-ECMO) utilized for the management of severe respiratory failure [21–23]. Of note, medical technologies have been finding an important role also in the so-called ‘long COVID,’ a debilitating and complex condition occurring in at least 10% of SARS-CoV-2 infections, consisting of a wide spectrum of symptoms (e.g. myalgic encephalomyelitis, chronic fatigue syndrome, postural orthostatic tachycardia syndrome, vasculitis, etc …) [23,24]. However, providing a complete and exhaustive overview of the existing literature on the various technological breakthroughs implemented during and post-pandemic is beyond the aim of this paper. * [25]

COVID-19 lessons are paving the way for new challenges for healthcare professionals. In this perspective, we firmly believe that, thanks to its strong multidisciplinary spirit, Expert Review of Medical Devices can play an essential role in disseminating scientific evidence in the field of medical technologies, helping the device research community grow even further, also shining a light on facilities’ availability and accessibility worldwide, through a dynamic and constructive discussion.

Declaration of interest

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Additional information

Funding

This paper was not funded.