Interreg (VΙ-A) IPA CBC “Greece-North Macedonia 2021-2027”

Patient safety is a major issue of global health concern, affecting all countries regardless of their level of development. It is an important parameter of quality in health care, as millions of patients worldwide suffer, or die each year from in-hospital infections.

The present PROJECT aims to use a global approach based on Artificial Intelligence (AI) in investigating the risks of in-hospital infections revealing the importance of in-hospital infections that occur and are detected by a fully customized hospital information system, on the cost of treatment.

The objectives of the present project are:

1. To carry out the present study are Machine Learning (M.L.) approaches on hospital information systems for the identification of compilations according to the patient database in Tertiary Hospitals both in Northern Greece and Former North Macedonia.
2. To extract the piloting methodological results of Greek Hospitals that are already encouraging (accuracy >90%) for the detection of in-hospital compilations and infections in specific departments, to both sides of consortium partners.
3. To develop and implement a global well-trained A.I. system that is able to reveal the risks of in-hospital infections and manage possible medication errors.
4. With the development of a novel inter-country cooperation, taking into account different practices and legislation, (in EU and out of EU), hospitals could identify errors due to:

- staff,

- equipment, and

- clinical patient management,

     5. Finally, introducing department- as well as patient-specific guidelines to be implemented.

     6. Disseminate the results as good-to-follow practices according to the piloting phase of the   project

1. Hospital Infections PANDEMIC

"...The 2019 US - CDC report emphasizes progress in combating antimicrobial resistance.

However, the CDC’s 2022 special report highlighting the impact of COVID-19 on antimicrobial resistance in the U.S. found that much of that progress was lost, in large part, due to the effects of the pandemic. The pandemic pushed healthcare facilities, health departments, and communities near their breaking points in 2020, making it very hard to maintain the progress in combating antimicrobial resistance.

... In 2013, CDC published the first Antimicrobial Resistance Threats Report, which sounded the alarm to the danger of antimicrobial resistance.

The 2013 and 2019 reports do not include viruses (e.g., HIV, influenza) or parasites.

The 2013 report stated that each year in the U.S. at least 2 million people got an antimicrobial-resistant infection, and at least 23,000 people died".

2.  Daily new confirmed COVID-19 cases and deaths

"COVID-19 Data Repository by the Center for Systems Science and Engineering at Johns Hopkins University"

3.  Opioid EPIDEMIC

"Opioid Crisis
Devastating consequences of the opioid epidemic include increases in opioid misuse and related overdoses, as well as the rising incidence of newborns experiencing withdrawal syndrome due to opioid use and misuse during pregnancy.

• Understanding the Scope of the Epidemic

Opioid Overdose
Opioid overdoses accounted for more than 42,000 deaths in 2016, more than any previous year on record. An estimated 40% of opioid overdose deaths involved a prescription opioid.

Why do Adults Misuse Prescription Drugs?
Policymakers can use this information from the National Survey on Drug Use and Health to help inform substance abuse prevention and treatment needs in their communities."

OUR AIM:

Machine Learning approaches are recognized as tools for constructing electronic health record-based risk models to select important predictors automatically from numerous features and face random or systematic errors, following the generalization of evidence-based approaches that are important to produce early prevention guidelines. With the use of machine learning in validated medical databases collecting data or early-stage information, dealing with new or unknown infections caused by viruses (as recently for Covid-19), the disease could be more accurately figured and predicted for better management of the fast-growing patient population.

Deep learning has been commonly used on image data and could be a useful tool to improve the differential diagnosis and proper manipulation of respiratory infections.

In this project, we introduce a fast-responding approach for guideline generation during the early stages of epidemics or pandemics with the induction of statistical models and A.I. algorithms translating the output of unsupervised learning of decision support Natural Language Processing applications.

The final output will be an online adaptive learning system that will generate predictive analytic models helping healthcare professionals to target high-risk populations and optimize prevention strategies using at the same time properly, all available healthcare resources.

Preliminary Results

(watch the video presentation)