(The Center Square) Lawmakers plan to interrogate the head of Eco Health Alliance, the group accused of conducting dangerous coronavirus research in Wuhan, China, just before the COVID-19 pandemic.
The Select Subcommittee on the Coronavirus Pandemic will hold a public hearing May 1 where Dr. Peter Daszak is expected to testify. Daszak is the president of Eco Health Alliance, a U.S. nonprofit health research company that used taxpayer-funded grants to conduct coronavirus research.
The lawmakers on the committee allege that newly obtained documents show Daszaks previous testimony misled the committee or misrepresented the facts.
These revelations undermine your credibility as well as every factual assertion you made during your transcribed interview, the letter said. The Committees have a right and an obligation to protect the integrity of their investigations, including the accuracy of testimony during a transcribed interview. We invite you to correct the record.
One of those obtained documents appears to show Daszak saying he plans to work with Wuhan researchers.
A federal grant database shows that Eco Health Alliance received millions of dollars from the federal government since 2014 to study coronaviruses that originate in animals and in some cases can transfer to humans, with an emphasis on China.
A key and highly disputed part of the inquiry is whether Eco Health Alliance research included making coronaviruses more dangerous.
Under former President Donald Trump, the federal National Institutes of Health cut all funding to the group in question over the controversy.
Under the Biden administration, funding has been restored, and NIH has emphatically stated that Eco Health Alliance did not play a role in the start of the pandemic.
Unfortunately, in the absence of a definitive answer, misinformation and disinformation are filling the void, which does more harm than good, NIH said in a 2021 statement. NIH wants to set the record straight on NIH-supported research to understand naturally occurring bat coronaviruses at the Wuhan Institute of Virology, funded through a subaward from NIH grantee EcoHealth Alliance. Analysis of published genomic data and other documents from the grantee demonstrate that the naturally occurring bat coronaviruses studied under the NIH grant are genetically far distant from SARS-CoV-2 and could not possibly have caused the COVID-19 pandemic. Any claims to the contrary are demonstrably false.
In 2022 and 2023, NIH awarded Eco Health Alliance a total of at least $1,230,594 to research the potential for future bat coronavirus emergence in Myanmar, Laos, and Vietnam.
The idea that the COVID-19 virus began in a Wuahn lab was once denounced as a conspiracy theory but has now gotten more widespread credibility.
The FBI announced last year after its investigation that COVID-19 most likely came from the Wuhan lab. That news came just after the Department of Energy also said the Wuhan lab was most likely the origin of COVID-19, though neither agency expressed a high degree of confidence in that theory.
Other groups have suggested it came from the Wuhan wet market, though no definitive answer has been settled on.
L.L. Bean has just added a third shift at its factory in Brunswick, Maine, in an attempt to keep up with demand for its iconic boot.
Orders have quadrupled in the past few years as the boots have become more popular among a younger, more urban crowd.
The company says it saw the trend coming and tried to prepare, but orders outpaced projections. They expect to sell 450,000 pairs of boots in 2014.
People hoping to have the boots in time for Christmas are likely going to be disappointed. The bootsare back ordered through February and even March.
"I've been told it's a good problem to have but I"m disappointed that customers not getting what they want as quickly as they want," said Senior Manufacturing Manager Royce Haines.
Customers like, Mary Clifford, tried to order boots on line, but they were back ordered until January.
"I was very surprised this is what they are known for and at Christmas time you can't get them when you need them," said Clifford.
People who do have boots are trying to capitalize on the shortage and are selling them on Ebay at a much higher cost.
L.L. Bean says it has hired dozens of new boot makers, but it takes up to six months to train someone to make a boot.
The company has also spent a million dollars on new equipment to try and keep pace with demand.
Some customers are having luck at the retail stores. They have a separate inventory, and while sizes are limited, those stores have boots on the shelves.
WASHINGTON (TND)
The U.S. Department of Justice on Tuesday announced its COVID-19 Fraud Enforcement Task Force (CFETF) seized more than $1.4 billion worth of stolen COVID relief funds during the last three years.
In a news release, Attorney General Merrick B. Garland said at least 3,500 defendants were charged with federal crimes, which toteld more than $2 billion. He also said more than 400 civil lawsuits were filed, which resulted in court judgements and settlements of over $100 million -- and at least $1.4 billion seized or forfeited.
The information was part of a 33-page report that highlighted the "widespread fraud."
At least $1.4B in stolen COVID relief funds seized, Department of Justice says
Deputy Attorney General Lisa Monaco shared similar sentiments.
Michael C. Galdo, who serves as the director of the Task Force, said there are still ongoing investigations into hundreds of cases involving identity thieves, transnational fraud and money laundering networks, large-dollar individual fraudsters, and businesses that helped facilitate the crimes.
"CFETF member agencies have improved their data analytics capabilities and are using these new skills to investigate fraud more efficiently and effectively," Galdo said.
The Task Force was established in 2021 to fight and prevent pandemic-related fraud.
Load more...
There have been 3385 new cases of Covid-19 reported in the past week. Photo: 123RF
There have been 3385 new cases of Covid-19 reported in the past week, and 13 further deaths attributed to the virus.
Of the new cases, 2109 were reinfections.
There were 145 cases in hospital as of midnight Sunday, the Ministry of Health said.
Canterbury was the worst affected region, followed by Waitemat, Capital and Coast and Counties Manukau.
Last Monday, 3399 new cases and seven deaths attributed to the virus were reported for the previous week, with 168 cases in hospital.
UK health technology assessor theNational Institute of Health and Care Excellence (NICE) has published final draft guidance recommending use of remdesivir for treating COVID-19 in eligible adults and children.
Marketed by US antivirals giant Gilead Sciences (Nasdaq: GILD) under the trade name Veklury, remdesivir was the first medicine for COVID-19 to receive full marketing authorization in 2022. However, as the pandemic receded, sales of the
This article is accessible to registered users, to continue reading please register for free. A free trial will give you access to exclusive features, interviews, round-ups and commentary from the sharpest minds in the pharmaceutical and biotechnology space for a week. If you are already a registered user pleaselogin. If your trial has come to an end, you cansubscribe here.
Try before you buy
All the news that moves the needle in pharma and biotech. Exclusive features, podcasts, interviews, data analyses and commentary from our global network of life sciences reporters. Receive The Pharma Letter daily news bulletin, free forever.
Become a subscriber
Unfettered access to industry-leading news, commentary and analysis in pharma and biotech. Updates from clinical trials, conferences, M&A, licensing, financing, regulation, patents & legal, executive appointments, commercial strategy and financial results. Daily roundup of key events in pharma and biotech. Monthly in-depth briefings on Boardroom appointments and M&A news. Choose from a cost-effective annual package or a flexible monthly subscription.
Last week, Direct Relief distributed 14,500 Covid-19 vaccines, donated by Pfizer Mexico, that will be administered to patients free of charge as part of an effort to protect people over the age of 12 in the State of Mexico from the Omicron variant.
Of the donation, 12,000 doses were transported to the Centro Nacional de Capacitacin y Adiestramiento, or CENCAD, of the Red Cross located in Toluca, State of Mexico. Another 2,500 vaccines will be administered in the Hospital Regional de Alta Especialidad of Ixtapaluca and via a vaccination campaign in the municipalities of Lerma and Huixquilucan.
The State of Mexico has recorded the second-largest number of confirmed Covid-19 cases in the country, only below Mexico City, and the state with the largest number of deaths caused by Covid-19.
Direct Relief is deeply thankful for the privilege to work with Pfizer Mexico on this important donation, which will benefit those in need in Mexico, said Eduardo Mendoza, National Director of Direct Relief in Mexico. With this donation, Direct Relief hopes to have an impact on health and health equality for the Mexican people.
Access is the cornerstone of healthy ecosystems, that is why we are thrilled to make this alliance with Direct Relief, which will allow us to get to a sector of the vulnerable population that needs better access paths to the treatment and vaccines that they need, said Constanza Losada, President and General Director of Pfizer Mexico.
As the United States eased into spring, only two states had increases or likely increases in coronavirus infections as of March 30, according to an update from the Centers for Disease Control and Prevention.
In addition, the CDC reported a decline or likely decline in coronavirus infections in 29 states and territories and called infections stable or uncertain in 19 areas. Even so, a national covid-19 forecast predicts up to 3,400 daily covid hospital admissions in late April.
All told, nearly 1.2 million U.S. residents have died of covid-19, according to the CDCs Covid Data Tracker.
To protect against a range of respiratory viruses, including the virus that causes covid-19, health experts recommend vaccination, along with good hygiene practices, including covering your nose and mouth when coughing or sneezing, washing your hands frequently with soap and water or using hand sanitizer, and regularly cleaning frequently touched surfaces like doorknobs, railings, desks and countertops.
Despite covids multiyear prevalence, not everyone has contracted it. Those who have not sometimes referred to as super-dodgers or novids represented nearly 1 in 4 U.S. adults and older teens by late 2022, according to various academic and medical groups citing CDC information. More recent data was not available as of press time, but its likely the novids number has declined since 2022.
This article is part of The Posts Big Number series, which takes a brief look at the statistical aspect of health issues. Additional information and relevant research are available through the hyperlinks.
NewsWatch 12 has the details.
MEDFORD, Ore. -- A group of lawsuits against Asante's health care operations are sharing a day in federal court.
Thirteen federal court cases have a shared proceeding this afternoon with the U.S. District Court in Medford.
A sampling of the cases shows claims of religious discrimination against Asante by health care workers who invoked religion as their rationale for refusing to get COVID-19 vaccinations, claiming their employment with Asante subsequently was suspended, causing wrongful employment termination for exercising their religious beliefs.
Eight of the cases list individual plaintiffs, and five of the civil cases have multiple plaintiffs, such asKather et al v. Asante Health System et al. For example, that case has 14 plaintiffs, including Michele Kather, Kourtney Selee, Alyssa Button, Justin Cirillo, Anna Drevenstedt, Ron Hittinger, Miles Kopish, Holly Martin, Tamara Rada, Jessica Stone, Michaela Begg, Myranda Miller, Ronda Osterberg and James Wilson. They list Asante Health System and Does 1 Through 50 as plaintiffs.
Their federal lawsuit seeks "damages for religious discrimintation in violation of Title VII ... and aiding and abetting religious discrimination."
A member of the Emergency Department staff of Asante Three Rivers Medical Center moves toward a patients room.
Today's court conference involves Asante Health System, Asante Rogue Regional Medical Center and Asante Three Rivers Medical Center as follow:
Salem-based attorney Ray Hacke is legal counsel for plaintiffs in three cases, including Kather.He said he believes all of today's Asante cases are COVID-19 vaccine related.
Hacke said Asante claims it accommodated employees who'd cited religion to decline COVID-19 vaccination by putting them on unpaid leave from work. He says, "Putting everybody on unpaid leave ... left people to choose between their faith and their employment."
A staff member of the Emergency Department at Asante Three Rivers Medical Center works at a computer terminal.
He said federal law is clear that employers should not penalize people who identify and articulate their religious beliefs as rationale for an allowed exemption from vaccination.He said, for example, Ronda Osterberg specifically spelled out her faith basis as rationale to Asante for declining COVID-19 vaccination, "then was fired outright."
Hacke said the United States Court of Appeals for the Ninth Circuit, with jurisdiction over Oregon, last year ruled that an employee's religious observance is a minor adjustment for employers to accommodate.
His three cases are among today's 13 cases he says are getting consolidated by the federal district court in Medford, including his largest plaintiffs case with 47 plaintiffs.
Hacke said he believes all 13 of today's cases involve Asante employee dismissals for refusing COVID-19 vaccine, and most of them involve religious discrimination claims for people who invoked their religion as a protected right for declining the vaccine, including Asante staff chaplain James Wilson. Hacke says a few cases might involve staff members who invoked protections in the federal Americans with Disabilities Act for medical exemptions from vaccination. His religious discrimination cases involve claims from more than 60 plaintiffs.
Hacke said one of his filings is a potential class-action case. He notes that with class action cases, instead of needing to prove all plaintiffs' cases, then legal counsel need to prove one case or a sampling of cases that are representative of other plaintiffs.He said Asante is resisting a class action which, "could involve a substantial amount of money."
Hacke said his three pending cases leave damages for court to determine. He said some of his clients have mitigated their damages, or financial losses from their claimed Asante employment dismissal, by going to work for Providence's Medford medical organization or returning to Asante after about six months, without vaccination. Hacke says Asante's reinstatement of those plaintiffs/employees and accommodating their religious exemptions later appears to undermine Asante's position of dismissing them in the first place.
Hacke said his cases leave the court to determine adequate compensation for distress, depression, anxiety and punitive damages because they involve factors that are unmeasurable, with real costs and losses somewhat determinable.
He said the federal court is coordinating these 13 cases to process them together through the judicial system.Hacke said other similar cases are underway in other places in Oregon.
Hacke said this way the ruling and the court processing of the cases should be consistent.
He said U.S. District courts in Oregon have 120 COVID-19 vaccination cases.
A staff member in the Critical Care Unit at Asante Three Rivers Medical Center treats a patient. Due to the influx of critically-ill patients suffering from COVID-19, staff had to add beds to some rooms.
He has three cases against Asante, with other federal cases pending against Rogue Community Health and another party.
Hacke explains that Title VII federal statute claims have "burden shifting" framework. He said first, plaintiffs must demonstrate their religious beliefs caused conflict. Second, he says the employee/plaintiff must show they informed the employer about the conflict. Third, Hacke said then-plaintiffs must show the employer in question did not accommodate the employee or the employer took adverse action against the employee/s.
Hacke says then the employer must show either undue hardship prevented reasonable accommodation for the employee or show that accommodation was made, which Asante says unpaid leave provides.
Hacke says unpaid leave is same as employment termination, and he says in the "burden (of proof) shifting," then the next step is plaintiff showing "pretext," or demonstrating the employer's case is inaccurate or its action is a not an accommodation.
Hacke expects the consistent, consolidated treatment by the federal court will have the cases advance or end as one. He says besides representing a plaintiff who was Asante's chaplain, his clients include a security guard and a person who worked remotely.
An isolation room at Asante Three Rivers Medical Center displayed a green check, indicating the room was occupied by a patient not suffering from COVID-19, and no special precautions were needed when entering. The colored sphere above the door is used to indicate negative pressure, ensuring airflow is into the room, not out of it.
Like most countries, Kyrgyzstans healthcare service delivery has been severely affected by the COVID-19 pandemic. Among others, the Global Fund to Fight AIDS, Tuberculosis and Malaria (GF), Gavi the Vaccine Alliance, Joint United Nations Programme on HIV/AIDS (UNAIDS), United Nations Development Programme (UNDP), United Nations Children's Fund (UNICEF), United Nations Population Fund (UNFPA), World Bank (WB) and World Health Organization (WHO) coordinated their actions to address various critical health issues identified by the government, such as reinforcing workforce capacity, increasing vaccination coverage and maintaining essential health services. Collaboration was enhanced through stakeholder consultations, high-level policy meetings, joint planning and resource mobilization. This collaboration aligns with commitments made by the agencies under the Global Action Plan for Healthy Lives and Wellbeing for All (SDG3 GAP) Action Plan for Healthy Lives and Wellbeing for All (SDG3 GAP). Activities undertaken during the pandemic have strengthened the country's primary healthcare (PHC) capabilities and its preparedness to respond to future crises.
By the end of 2021, Kyrgyzstans COVID-19 vaccination coverage remained considerably below the WHO global target of 70%, primarily due to vaccine hesitancy, political instability, and reorganization within the Ministry of Health (MoH). Gavi, UNDP, UNICEF, WB and WHO assisted the Government in securing vaccines through COVID-19 COVID-19 Vaccines Global Access (COVAX), direct procurement and donations. They facilitated national vaccine roll-out plans, strengthened supply chains and expanded vaccination points, including PHC clinics. In addition to COVID-19 vaccines, Gavi and WHO supported the MoH to rapidly expand mobile immunization teams for routine vaccinations. These teams conducted six rounds of visits, covering 88% of remote and hard-to-reach areas, reaching over 170,000 individuals. UNICEF supported improving cold-chain infrastructure and increased access to specialized refrigeration equipment at vaccine points from 49% to 85% nationwide (1). They also reached parents of 8000 under-vaccinated children and empowered 3000 members of 952 village health committees for advocacy on immunization.
A 62-year-old man receives a dose of flu vaccine at the Nijnaya Serafimovka Center for the Elderlyand Disabled in Chuy Oblast, Kyrgyzstan, 25 November 2022. WHO / Arete / Maxime Fossat
By early 2020, HIV and TB service delivery, alongside many other essential health services fell behind due to the diversion of resources to COVID-19. To support the Government response, GF, UNAIDS and UNDP assisted in reaching TB patients and people living with HIV by introducing mobile teams and utilizing messaging applications for online counseling and referrals. UNAIDS-supported peer consultants remained in contact with people living with HIV, delivering medicine to their homes so they could stay on treatment during the lockdown, distributing food packages and providing psychological support.
"Working as a peer consultant has given me an opportunity to help people to overcome their problems, many of which I have come across myself in the past,"said Toktonalieva, a peer consultant.
Complementing those efforts, UNFPA, UNICEF and WHO collaborated with the Government to conduct health surveys which revealed reduced availability of essential sexual, reproductive, maternal, neonatal and child health services due to restricted movement during lockdowns. UNICEF and the MoH piloted a project in select communities that introduced a model for postpartum home visit services coupled with teleconsultations to provide personalized care, information on childhood illnesses, nutrition and early childhood development milestones.
"Every day on TV health officials reported about the number of infected people and deaths. During this period, I gave birth to my long-awaited baby girl. On the second day after being discharged from the hospital, our family doctor knocked on our door. She stood at a distance and asked to see me and the baby, examined the skin, asked me to breastfeed, while explaining the rules of attachment and position to the breast. She also mentioned the dangerous signs of diseases in newborns and mothers after childbirth. The home visit of our family doctor became a real source of support and confidence," expressed the mother of a five-day-old newborn.
A health professional swabs a woman, testing her for COVID-19, September 2020. WHO
In tandem, UNFPA and WHO assisted the Government in the development of a clinical protocol for the management of sexual violence cases and trained 412 PHC providers, as gender-based violence peaked during the pandemic. A healthcare provider from the southern region of Kyrgyzstan noted that she acquired improved skills in providing immediate care for rape victims, and a better understanding of the necessary steps required for referring survivors to a service provider.
She added, "It is crucial for women to have privacy during counseling, particularly because many women report feeling ashamed, embarrassed, and blame themselves for the violence they experience. Good counseling services [by us] enhanced the supportive process by validating their experiences and feelings, being sensitive to their situations, providing a non-judgmental attitude, and ensuring a safe environment."
Following the trainings, two methodological courses on sexual and gender-based violence were developed for teachers and healthcare providers. These courses were institutionalized in Kyrgyz State Medical Institute on Continuous Education, Kyrgyz Slavonic University, Issyk-Kul and Osh State Universities for 24 mandatory credit hours in undergraduate and post-graduate degrees.
Challenges in collaboration stemmed mainly from funding shortages and strained health systems. To address the anticipated US$ 500 million financing gap resulting from trade and mobility disruptions, the Government requested an additional US$50 million funding from bilateral and multilateral development partners (2). Subsequently, the development partners, including some SDG3 GAP partners, pledged US$ 45 million to bolster the health sector's response to COVID-19. These funds primarily target the acquisition of medical equipment, pharmaceuticals, training initiatives, and laboratory services. Some of these contributions are allocated over 23 years. Valuable lessons were learned from these collaborative efforts that emphasized the importance of coordinated responses, adaptability and robust partnerships in addressing complex health challenges.
(1) Country Office annual report 2022 - Kyrgyzstan. https://www.unicef.org/media/136016/file/Kyrgyzstan-2022-COAR.pdf
(2) World Health Organization. (n.d.). Health Financing in Kyrgyzstan: Obstacles and opportunities in the response to COVID-19. World Health Organization. https://www.who.int/europe/publications/i/item/WHO-EURO-2021-2604-42360-58654
The modified lumped-parameter model consists of a system of ordinary differential equations (ODEs) that needs to be numerically solved to allow the computation of different model outputs of clinical interest. We calibrated the model to fit some clinical data of patients hospitalized for severe COVID-19-related pneumonia in the Internal Medicine ward of L. Sacco Hospital in Milan, Italy, between March and April 2020. We analysed the statistical reliability of the model outputs for each successful calibration by means of uncertainty intervals and, finally, we performed a statistical analysis on clinical data or model outputs by means of hypothesis tests to highlight the impairments of the cardiocirculatory system associated with COVID-19 pneumonia.
We identified four groups of quantities, taken from the dataset or obtained as an output of the calibrated model:
The clinical data used for the model calibration, obtained from clinical measurements and referring to physical quantities (PQ1), as, for example, the maximal left atrial volume (LAVmax) and the systolic systemic pressure (SAPmax);
The inputs of the model (heart rate HR and body surface area BSA) and of the calibration procedure (right ventricular fractional area change RVFAC and tricuspid annular plane systolic excursion TAPSE), provided by other clinical measurements;
The parameters of the model (e.g. resistances and compliances) determined through a calibration procedure, from now on referred to as calibrated parameters;
The outputs of the numerical simulation of the model (e.g. flow rate and mean pressure), from now on referred to as model outputs. Some of them (MO1) referred to physical quantities (PQ1) that were also measured (clinical data), for example, LAVmax and SAPmax. Other model outputs (MO2) referred to physical quantities (PQ2) that were not measured but quantified only by means of the computational model. Examples of the latter are the mean left atrial pressure (LAPmean) and indexed right ventricular end diastolic volume (RVI-EDV). The complete list of PQ1 and PQ2 is reported in Supplementary Table 1.
We remark that the indexed value of volumes of a patient can be computed dividing the volumes by the BSA of that patient (Supplementary Table 2). In what follows, an I- that precedes a subscript of a volume means that the volume is indexed (for example, LVI-EDV is the indexed left ventricular end diastolic volume).
For the sake of clarity, we reported in Fig.1 the diagram flowchart of the followed procedure that is described in detail in what follows.
Diagram flowchart of the procedure used in this study. Top: calibration; mid: statistical analysis of measured physical quantities; bottom: statistical analysis of computed physical quantities. Calibration: the mathematical model required as inputs HR and BSA of a specific patient. The model computed MO1 using an initial setting of parameters (that could need to be calibrated, so they are highlighted in red). If MO1 were close enough to the clinical data the model was considered calibrated (the parameters are highlighted in green); if not, the calibration method was iteratively applied to the parameters using RVFAC and TAPSE as inputs. If the parameters were not modified the calibration failed; if not, MO1 were recomputed by using the new setting of parameters and the previous steps were repeated. Statistical analysis 1: we performed hypothesis tests on clinical data (test I). Statistical analysis 2: HR and BSA were used as inputs of the calibrated model for every patient with a successful calibration, the model computed the MO2 and we checked the statistical reliability of MO2. We collected the reliable MO2 from every patient and we performed hypothesis tests on the reliable MO2 of all the patients (test II).
The dataset consists of (58) patients, who all required oxygen supplementation but none of them was on mechanical ventilation. Of such patients, only (29) were calibrated according to point (iii) above (see Calibration subsection below) ((56pm 18) years). Such patients did not present symptoms or signs of heart failure or substantial structural cardiac disease; (10) out of (29) were older than (64) years; (6) patients had arterial hypertension, (1) had diabetes and (4) showed the association of hypertension and diabetes (Supplementary Table 2).
The echocardiography of each patient was performed early after the admission to the hospital. Examinations were performed at bedside using a Philips CX-50 portable device by expert operators. Measures were defined according to the latest European and American Echocardiography Society guidelines18,19.
Each patient provided consent to use his/her data for observational studies. The institutional board has approved the study with protocol number 16088/2020.
The cardiovascular system was studied by means of a lumped-parameter (0D) mathematical model that splits the system into compartments (e.g. right atrium, systemic arteries/veins) and, for each of them, the time evolution of model outputs (pressures, flow rates and cardiac volumes) is modelled by a system of ODEs20,21. The lumped-parameter model is described through an electrical circuit analogy: the current represents the blood flow through vessels and valves; the electric potential the blood pressure; the electric resistance plays the role of the resistance to blood flow; the capacitance represents the vessel compliance; the inductance the blood inertia; the increase in elastance the cardiac contractility.
There are different possible choices and number of compartments, depending on the purpose of the study, for the construction of a lumped-parameter model (e.g. Refs.16,17,22,23). We considered the computational model introduced in Ref.17, wherein the four heart chambers, the systemic and pulmonary circulations, with their arterial and venous compartments were included, and we substituted the 3D left ventricle with a 0D component (as in Ref.16) and we added two new compartments accounting for systemic and pulmonary capillaries. The pulmonary capillary circulation was also split in two compartments accounting for oxygenated and non-oxygenated capillaries (Fig.2).
Lumped-parameter cardiocirculatory model. The unknown pressures and flow rates are in red and blue, respectively, whereas the model parameters are in black. Notice in the green boxes the new compartments with respect to16 featuring this work.
The system of ODEs associated with the lumped-parameter model is formed by the equations representing continuity of flow rates at nodes and of pressures in the compartments, and its numerical solution allows to compute several model outputs as functions of time: the left and right atrial and ventricular volumes (({V}_{{text{LA}}}), ({V}_{{text{LV}}}), ({V}_{{text{RA}}}) and ({V}_{{text{RV}}})), the systemic and pulmonary arterial, capillary and venous pressures (({p}_{{text{AR}}}^{{text{SYS}}}), ({p}_{{text{C}}}^{{text{SYS}}}), ({p}_{{text{VEN}}}^{{text{SYS}}}), ({p}_{{text{AR}}}^{{text{PUL}}}), ({p}_{{text{C}}}^{{text{PUL}}}) and ({p}_{{text{VEN}}}^{{text{PUL}}})), the systemic and pulmonary arterial and venous blood fluxes (({Q}_{{text{AR}}}^{{text{SYS}}}), ({Q}_{{text{VEN}}}^{{text{SYS}}}), ({Q}_{{text{AR}}}^{{text{PUL}}}) and ({Q}_{{text{VEN}}}^{{text{PUL}}})).
Starting from these functions, it is possible to compute the pressures of the four cardiac chambers (({p}_{{text{LA}}}), ({p}_{{text{LV}}}), ({p}_{{text{RA}}}) and ({p}_{{text{RV}}})), the blood fluxes through the valves (({Q}_{{text{MV}}}), ({Q}_{{text{AV}}}), ({Q}_{{text{TV}}}) and ({Q}_{{text{PV}}})), through the systemic capillaries (({Q}_{{text{C}}}^{{text{SYS}}})) and through oxygenated and non-oxygenated pulmonary capillaries (({Q}_{{text{C}}}^{{text{PUL}}}) and ({Q}_{{text{SH}}})), and all the model outputs referring to PQ1 and PQ2 (Supplementary Table S1).
We considered reference values of the parameters (such as resistances and compliances) such that all the model outputs were in the reference healthy ranges of the corresponding physical quantities taken from the literature7,18,19,24 for an ideal individual with HR equal to (80) bpm (beats per minute) and BSA equal to (1.79) m2 (Supplementary Table S3). We did not consider model outputs computed starting from the flow rates, because they are not uniquely defined depending on the tract of the compartment where they are measured, from ({p}_{{text{C}}}^{{text{SYS}}}), due to the heterogeneity of the pressures of systemic capillaries among tissues, and from ({p}_{{text{VEN}}}^{{text{SYS}}}), even if we recovered the value of central venous pressure, that coincides with the right atrial pressure24.
We reported the system of ODEs associated with the lumped-parameter model in Supplementary Equations S1. The lumped-parameter model was numerically discretized by means of Dormand-Prince method25 (adaptive stepsize RungeKutta) which was implemented in Python using the Jax library26.
The lumped-parameter model was characterized by parameters representing the functional properties of the compartments (e.g. resistances). To properly select such values for a specific compartment and patient, a calibration procedure was needed27,28.
We chose a priori the cardiac timings and the resistance of oxygenated pulmonary capillaries (({{text{R}}}_{{text{C}}}^{{text{PUL}}})) equal to the associated reference values. In particular, we fixed ({{text{R}}}_{{text{C}}}^{{text{PUL}}}) to avoid modelling micro-thrombosis because of its possible increase. For the remaining parameters, the calibration of the model relied on the method we presented in Ref.27, that is aimed to reduce the sum of squared relative errors between the model outputs MO1 and clinical data, modifying the parameters of the model in suitable bounded intervals ({{text{I}}}_{{text{i}}}), for (i=1,dots ,{N}_{{text{p}}}), where ({N}_{{text{p}}}) is the number of parameters, independent of the patient, built starting from the reference values of parameters mentioned before (Supplementary Table S3). Specifically, we chose to calibrate those parameters among the latter according to a sensitivity analysis estimating the absolute correlation coefficients between parameters and model outputs (Supplementary Table S4). We calibrated only the parameters featuring at least one absolute correlation coefficient greater than (0.1) that was associated to provided clinical data. To reproduce the blunted hypoxic pulmonary vasoconstriction condition, the resistance of non-oxygenated pulmonary capillaries (({{text{R}}}_{{text{SH}}})) could decrease in such a way that the shunt fraction could reach values up to (70%) in the worst-case scenario. The list of amendable parameters varies between different patients according to the different clinical data provided.
The calibration was based on clinical measurements of COVID-19 patients that were provided by L. Sacco Hospital in Milan and referred to HR and BSA, which were used as inputs for the lumped-parameter model, RVFAC and TAPSE, which determined the bounded interval ({{text{I}}}_{overline{{text{i}}} }) used during the calibration, with (overline{i }) the index referring to the right ventricular active elastance, and the clinical data, given by a subset of the pressures and volumes involved in the cardiac circulation (Supplementary Table S2).
To provide further mathematical details, we indicate with (mathbf{p}) a configuration of parameters of the cardiocirculatory model. The calibration method aimed to find the configuration of parameters ({overline{mathbf{p}} }^{{text{j}}}) which minimized the loss function for the specific patient (j), that reads:
$$L^{{text{j}}} left( {mathbf{p}} right) = sumlimits_{{{text{l}} = 1}}^{{{text{N}}^{{text{j}}} }} {left( {frac{{q_{{{text{m}}_{{text{j}}} left( {text{l}} right)}}^{{text{j}}} left( {mathbf{p}} right) - d_{{text{l}}}^{{text{j}}} }}{{d_{{text{l}}}^{{text{j}}} }}} right)^{2} } ,$$
(1)
where ({N}^{{text{j}}}) is the number of available echographic clinical data for patient (j), ({d}_{{text{l}}}^{{text{j}}}) is the value of the l-th clinical data of patient (j) (Supplementary Table S2) and ({q}_{{{text{m}}}_{{text{j}}}left({text{l}}right)}^{{text{j}}}) is the value of the model output related to the l-th clinical data of patient (j). The index (m) of ({q}_{{text{m}}}^{{text{j}}}) lies in ({1,dots ,{N}_{{text{q}}}}) where ({N}_{{text{q}}}) is the number of both MO1 and MO2. We considered the model calibrated for a specific patient if the loss function was below ({10}^{-3}). Notice that, for some patients, the calibration procedure could fail, if, for example, it reaches the minimum of the loss function that is above the required threshold.
Moreover, to improve the robustness of the calibration procedure, we repeated, for every patient, the calibration three times, with different initial configurations of parameters, and we considered the calibrated setting of parameters that returned the lowest loss function. As anticipated above, only (29) out of (58) patients were successfully calibrated. We noticed that by performing (4) times the calibration procedure the number of calibrated patients was still equal to (29), precisely as after (3) calibrations.
The loss function (1) was minimized by the Quasi-Newton method L-BFGS-B29 implemented in Scipy by computing its gradient by means of automatic differentiation (reverse mode gradient) included in the library Jax26.
For every patient (j) calibrated with a loss function below ({10}^{-3}), a configuration of parameters ({overline{mathbf{p}} }^{{text{j}}}) was at disposal. The loss function was computed using the clinical data provided by L. Sacco Hospital, which were related to measurement errors (Supplementary Table S1), that also affected the uncertainty of the model outputs ({mathbf{q}}^{{text{j}}}). We needed to determine, for every patient, if the related model outputs were reliable or not, so we proceeded along two steps:
Build a sample of candidate model outputs ({mathbf{q}}^{{text{j}},{text{k}}}) for (k=1,...,n) ((n) was (100));
Determine, by employing a simple statistical analysis, whether the mean of the model outputs was reliable.
Regarding step 1, for every provided clinical data ({d}_{{text{l}}}^{{text{j}}}) of patient (j), we built an interval ({{text{M}}}_{{text{l}}}^{{text{j}}}) centred in the value of the clinical data with width equal to two times the measurement error (Supplementary Table S1). Then, we built the samples ({mathbf{q}}^{{text{j}},{text{k}}}) by following the subsequent procedure:
Choose a relative width (w) ((w) was (12.5%));
Build an interval centred at ({overline{p} }_{{text{i}}}^{{text{j}}}) and with width (2w{overline{p} }_{{text{i}}}^{{text{j}}}) for every (i=1,dots ,{N}_{{text{p}}}). If this interval is not included in the parameter interval ({{text{I}}}_{{text{i}}}) used for the calibration, then cut off its overflowing extremities.
Perturb every parameter of the calibrated patient sampling from a uniform distribution in the corresponding interval built at point b) thus obtaining ({p}_{{text{i}}}^{{text{j}}});
Run a simulation of the cardiocirculatory model with parameters ({mathbf{p}}^{{text{j}}});
Check if the model output ({q}_{{{text{m}}}_{{text{j}}}left({text{l}}right)}^{{text{j}}}) generated at point d) lie in the intervals ({{text{M}}}_{{text{l}}}^{{text{j}}}). If they do, save the new configuration of acceptable model outputs ({mathbf{q}}^{{text{j}}}), otherwise reject it;
Repeat from point c) until (n) iterations are performed;
Check if the acceptance ratio (ratio between the number of saved configurations and the number of iterations) is within ([0.1, 0.15]). If it does, repeat from point c) to e) until (n) configurations are accepted because at this step the sample size of candidate model outputs is small (with (n=100), the size is between (10) and (15)), otherwise increase or decrease (w) to retrieve the condition on the acceptance ratio, discard the previous configurations and repeat from point b).
Once the above procedure was concluded, we proceeded with step 2 by using the (n) samples of acceptable model outputs ({mathbf{q}}^{{text{j}},{text{k}}}) for (k = 1, dots , n) generated at the previous step, for every specific patient (j). If the standard deviation of the sample of a model output of patient (j) was lower than 5% of its mean, we considered the mean reliable and we used it for the hypothesis tests. In this way, for every model output we built a sample of accepted values (depending on the patient), where sample size depended on the considered model output.
Prediction intervals could have been used for this analysis, but, if the sample was not normally distributed, a link function would be needed to retrieve normality30. We checked, for every patient (j) and for every model output, if the sample of that model output was normally distributed by means of a chi-squared test. It turned out that the sample is not normally distributed for all patients. Thus, since we wanted to use the same statistical approach for every patient, we resorted to this heuristic approach based on standard deviation instead of prediction intervals.
If the sample mean, calculated over all patients, of a clinical data or MO2 (referring to physical quantities PQ1 and PQ2, respectively) fell inside the healthy range of the corresponding physical quantity7,18,19,24, we did not consider the physical quantity altered in association with COVID-19 infection, otherwise we performed hypothesis tests to check whether the mean was significantly (p-value below (0.01)) increased or decreased with respect to the healthy range to investigate the impairments of the cardiovascular system in association with COVID-19 infection. If the sample mean, calculated over all patients, was less than the lower bound of the healthy range, the null hypothesis was that the mean was greater or equal than the lower bound of the healthy range, whereas the alternative hypothesis was that the mean was smaller than the lower bound of the healthy range. If we accepted the null hypothesis, then the corresponding physical quantity was considered not altered in association with the infection of COVID-19; otherwise, we considered the physical quantity altered in association with COVID-19. If, instead, the sample mean was greater than the upper bound of the healthy range, we proceeded similarly.
For each clinical datum, we computed the mean and the standard deviation of its sample without resorting to the mathematical model. The sample sizes were large enough to use one-tailed z-tests (assuming the variance equal to the unbiased sample variance) comparing their means to the nearest bound of the healthy range (test I).
For every MO2 we computed the mean and the standard deviation of its sample. We performed a chi-squared test and not every sample was normally distributed, so we opted for one-tailed z-tests (assuming the variance equal to the unbiased sample variance) only if the sample had more than (24) elements comparing their means to the nearest bound of the healthy range (test II).
Notice that for group PQ1 the statistical analysis was carried out directly using the clinical data and not the MO1 values. Accordingly, the clinical data were used in a twofold way:
To statistically compare PQ1 clinical measures with healthy ranges independently of the application of the proposed lumped-parameter model (test I);
To calibrate the lumped-parameter model for the patients at hand thus allowing to obtain MO2 that are statistically compared with healthy ranges (test II).
