The Real-World Data (RWD) from the HUS Data Lake Provides Clinicians with Valuable Information about Treatment Outcomes

Hospital data lakes offer clinicians diverse real-world data (RWD) that supports their work. Dr. Heikki Ekroos, Chief Physician of Pulmonary Diseases at Helsinki University Hospital (HUS) Porvoo Hospital, presented real-world evidence (RWE) studies that were based on HUS data lakes data at Medaffcon’s EMMA client event.   

The Helsinki University Hospital (HUS) data lake contains a vast amount of healthcare data on the population in the Helsinki-Uusimaa Region. This data lake includes healthcare data of 3.5 million individuals and, among other things, 37 million patient records and 790 million laboratory results. 

Dr. Heikki Ekroos, Chief Physician of Pulmonary Diseases at Helsinki University Hospital (HUS) Porvoo Hospital, is involved in studies that investigate the relationship between smoking and surgical complications and their associated costs. Other ongoing studies explore the profiles of sleep apnea patients and treatment outcomes for lung cancer patients. 

All data on lung cancer patients is extracted from the HUS data lake 

The lung cancer study includes all data from patients diagnosed between 2013 and 2023. The data is extracted from the Helsinki University Hospital (HUS) data lake. This study includes all available information, such as healthcare visits, treatments, genetics, pathology, and laboratory tests. 

The results provide information, among other things, on treatment outcomes and will help clinicians make evidence-based treatment decisions. 

-Previously, we didn’t have any data on lung cancer treatment outcomes. Yet, we are constantly taking actions and administering expensive medications. It’s essential to know what happens to patients. Now we get real-world data (RWD) quite rapidly, says Dr. Heikki Ekroos, Chief Physician of Pulmonary Diseases at Helsinki University Hospital (HUS) Porvoo Hospital. 

The data lake provides information on smoking and surgical complications  

A registry study on the relationship between smoking and surgical complications was published in March 2023. Its data consisted of all surgeries performed in the Helsinki University Hospital (HUS) area between 2015 and 2019.  

The study showed that current and former smokers have a significantly increased risk of experiencing complications after surgery. This study was also a great example of the possibilities that machine learning and hospital data lakes provide for research. 

– It was important for me that we obtained scientific results on the effects of smoking on surgical recovery. Now we have a strong argument for why it’s worth quitting smoking before surgery, says Dr. Heikki Ekroos, Chief Physician of Pulmonary Diseases at HUS Porvoo Hospital, who led the study. 

Machine learning and artificial intelligence assisted in real-world evidence (RWE) research  

The dataset for the study covered about a million surgeries. Initially, clinicians reviewed 20,000 patient records and classified patients as smokers, quitters, or non-smokers. This classification was used to teach a machine learning algorithm.  

By using machine learning and artificial intelligence, smoking data for over a million surgeries was determined. The algorithm distinguished smokers, quitters, and non-smokers from patient records. Physicians had predefined the complications being sought. Ultimately, 158,638 surgeries were left for AI analysis. 

According to Ekroos, Medaffcon’s expertise was essential for the project and helped define how to use the data. 

-At the beginning of the project, it is necessary to define what data is being sought. Medaffcon also helped integrate machine learning into the study. Without machine learning, handling such a large amount of data would have been impossible, Dr. Heikki Ekroos, Chief Physician of Pulmonary Diseases at Helsinki University Hospital (HUS) Porvoo Hospital explains.