COVID-19 … What are drugs and strategies now?

From February 2019 the World faces the Covid19 pandemic. The data in our possession are still insufficient to effectively combat this pathology. The gold standard for diagnosis remains molecular testing, while clinical and instrumental and serological diagnostics are highly nonspecific leading to a slowdown in the battle against covid19.[3] Can Artificial Intelligence (AI) and Machine Learning (ML) help us? The use of large databases to cross-reference data to stratify the diagnostic scores, to quickly differentiate a critical Covid-19 patient from a non-critical one is the challenge of the future. All to achieve better management of resources in the field and a more effective therapeutic approach.[2].

To the Editor,

Since February 2019 Europe has been faced with the COVID-19 infection. WHO in March announced the global emergency, and, to today, the total COVID-19 cases in Europe are 24,917,775, and the deaths 548,256 (1). At present, the question is: what are the pharmacological and diagnostic evidences to manage this second wave in a better way? Can Artificial Intelligence (AI) and Machine Learning (ML) help us? (2) It is essential, through algorithms obtained with the technologies in our possession, to stratify the risk and create clinical diagnosis scores with real support of laboratory and instrumental diagnostics, all in order to reach curative therapy in an effective way.

The patients’ clinic continues to be nonspecific. In contrast, some patients have specific characteristic symptoms, such as ageusia and anosmia (3). Comorbidities are often present (hypertension, diabetes, obesity, cerebrovascular accidents and oncological pathologies) (3).

Molecular testing represents the gold standard for diagnosis (3,4).

Laboratory test identifies an increase in C-Reactive Protein (C-PR), lymphopenia and specific markers may be altered, such as cardiac, renal and hepatic ones. The coagulation cascade is also affected, and microthrombi formation into pulmonary bed are described associated with increase in D-Dimer due to alteration of the coagulation process (3,4). This is a form of Acute Respiratory Distress Syndrome (ARDS) visible on the CT scan (3-5) with characteristic ground-glass opacities.

The Lung Ultra Sound (LUS) (3,4) can be used to follow the patient’s trajectory and reduce ionizing radiation in terms of chest X-ray and CT scan, with less healthcare worker exposition to the COVID-19 area. The LUS (4) identifies an ARDS-like pattern, B lines converging even to the “white-lung” but alternating with savings areas (spared areas) (3,4). With the use of AI and ML, it is possible to compare the numerous diagnoses to identify similarities or pathognomonic characteristics of the pathology, with the aim of drawing clear scores. The use of LUS also guarantees, thanks to ML, a study of the pathology even at a distance and in real time by an experienced operator, even in absence from the patient’s bed, thus avoiding further contagion problems (Tab. 1).

Table 1.

Diagnosis
Laboratory diagnosis						Instrumental diagnosis
Molecular		Sierological		Biochemical		TC		LUS
PROS	CONS	PROS	CONS	PROS	CONS	PROS	CONS	PROS	CONS
Diagnostic for COVID-19	NO Sensitive	Sensitive	NO Specific	Pathology and complications monitoring	NO diagnostic for COVID-19	Sensitive	Not Very Specific	NO X-Ray	Operator Sensitive
Fast	Annoying	Monitoring for Plasma		Diagnosis sepsis, MOF	NO Specific	Ground glass shadow and effusion and pulmonary infiltrating shadow	X-Ray	Pathology monitoring: Blines (multifocal, discrete, or confluent)	NO early diagnosis
						NO Operator Sensitive	High costs	Low costs	Not sensitive and specific for COVID-19
						Fast	Repeatable at the patient’s bed

However, at the moment, there is no specific treatment for COVID-19 patients. Encouraging results come from Corticosteroids, Heparin (3,5).

Dexamethasone 6mg once a day for ten days seems to reduce the levels of cytokine storm in the acute phase of COVID pathology.

Heparin and antiplatelet find their rationale in the management of micro-thromboembolic complications (12,13). In addition, many patients require supportive care: Paracetamol (5), up to the support of vital parameters with amine (5), intubation and mechanical ventilation in intensive care (5).

Once again, AI and ML can support us in the processing of large numbers. Identifying in advance future critical patients, and also non-critical patients, means treating them better. Being able to precisely elaborate therapeutic strategies means being able to manage certain categories of patients at home, with a better use of resources and consequent savings for the health system. This was also shown by a retrospective study highlighting how the aid of technology and AI can be useful in differentiating critical and non-critical COVID-19 patients (2,6)

Future strategies involve processing big data through Biobanks, or the foundations of consortia and ad hoc committees.

In conclusion, in our opinion, prevention, consistent and rapid diagnosis and a shared therapeutic approach are certainly a way forward to first contain and subsequently eradicate coronavirus.

Pursuing this difficult and complex path, also considering the large numbers we are dealing with, new technologies could provide useful and valid supports in daily clinical practice in this new phase of the present health emergency (6).