Comprehensive Cancer Database May Enhance Personalized Therapy

Leila Tchelebi, MD

Northwell Health

Integrating healthcare data from several disparate electronic medical records (EMRs) within a large healthcare system may help streamline patient data, consolidate information, and track survival outcomes across networks, according to findings that Leila Tchelebi, MD, and colleagues presented at the 2025 American Society for Radiation Oncology Annual Meeting.

CancerNetwork® spoke with Tchelebi, a radiation oncologist at Northwell Health, to learn more about her group’s efforts to develop and validate a unified cancer database across 14 institutions that may facilitate multi-center oncology research and provide a foundation for data-driven clinical practice. She elaborated on how this novel system may help inform individualized treatment strategies based on specific patient data. Furthermore, she emphasized spreading the system’s use on a greater scale at institutions across the US as a critical next step for research.

“I hope others will take away the value of this type of data consolidation. In the world of artificial intelligence [AI], I think tools like this are critically important to help us streamline various processes within the health system, [and] consolidating data into 1 easy-to-access place so that it can be used for enhanced patient care,” Tchelebi stated. “It is wonderful, and we have shown that it can be done. It's just a matter of other sites adopting this and utilizing it to streamline their own processes.”

CancerNetwork: What was the rationale for assessing this unified cancer database across 14 treatment facilities? Where did the idea for this comprehensive database originate?

Tchelebi: One of the troubles in medicine is that information that can be essential for patient care is distributed amongst different medical records that do not necessarily interface with one another, which results in a lot of extra time and effort for clinicians to chase down the information in the various medical records. This becomes particularly problematic at a large health system such as Northwell, where we have multiple hospitals on the inpatient and the outpatient side, different labs, and different places where patients can get imaging studies done. If you need to collect all the data for an individual patient from these various places, faxes get involved, and [you] must collect [them]. You cannot always [receive] the records electronically, and it becomes a big mess and a challenge for multiple areas of care, not only the care in the clinic, but as far as research you might want to conduct. The cancer registry collects information on patients that is required to be submitted nationally to state and government registries, so there are many reasons why all of a patient’s relevant medical history needs to be consolidated into 1 easy-to-find place, if possible.

The purpose of this research was to see if it was feasible. There is a database that has been created; it’s called Oncora. [It] collects data from these disparate EMRs. The question was, can we utilize those data in an effective means across our cancer network? That was our goal: to see if it was feasible, to see if this could be done, and if it would be useful. We found that, in fact, it truly is.

Logistically, how does this cancer database operate? What methods make this database stand out from other systems?

The database is available at a particular website that is institution specific. Northwell has an Oncora interface that we can access. There are [several] data elements already there; those get imported into the database, and then you can extract them based on the inquiry that you input. If there are variables that are missing—for example, another research project that we presented at ASTRO required some lab values that were not already existing in the database—we are able to contact the creators of the database, and then they can work with their team to extract those data. Truly, it seems to be limitless as far as what data can be imported. If it is not already there, it can be added.

What did findings from the study show regarding the efficacy of the development and implementation of this integrated database? What impact did the database have on patient care?

This was just the tip of the iceberg....to make sure that it is feasible and that these data can actually be consolidated. We were able to take over 100,000 cancer cases across our 14 facilities, totaling over 200,000 patient records. We consolidated these so that we would be able to implement survival tracking across the cancer registry that is available. We were able to integrate the different treatments that the patients had received between chemotherapy, radiation therapy, and surgery all into 1 place, so that for 1 individual patient, you [could] go to that record and track exactly what treatments were rendered, rather than having to look in multiple different places.

We were able to incorporate genomic profiles from laboratories for those individual patients; this could be used to see what genetic mutations the patients might have that might impact future cancers down the road. Then, the records were consolidated; by collecting data from the various databases with different EMRs for each patient across the different databases, we were able to consolidate and significantly decrease the number of records so that for 1 individual patient, instead of having 5 or 6 different EMRs across the system, it was now consolidated into 1 data set within this database. [This makes] it easy for various providers, again, to see what treatments the patients had, what their lab values are, and what genetic mutations they might have all in 1 place.

How might a system like this bolster collaboration across different academic, teaching, and community hospitals, as well other institutions?

The beauty about this is that it helps with multiple different aspects of healthcare. From a clinical standpoint, physicians across different sites within the Health Network— not only within the health system, but [even] outside the system—if these data are shared, [they] can retrieve all the relevant information regarding a patient for ease of decision-making clinically. I [can] look at my patient and see what chemotherapy they have received already, what radiation therapy they might have received at another facility that might not have the same EMR; perhaps it was an inpatient facility, and I am working at an outpatient site. Clinically, it has been [quite] helpful for the cancer registry that is required, as I mentioned previously, to collect these data and submit them to the state. All the information regarding the patient’s cancer stage, treatments rendered, and survival outcomes can be accessed in 1 place.

Then, for research, if we are looking to do a quality assurance project across the network [and] see how many patients perhaps did not complete therapy, we can have all the relevant clinical information in 1 easy-to-access database. Then, all that information can be exported to something like an Excel spreadsheet, which can then be utilized to much more easily conduct clinical studies where thousands of patient cases can be imported into the spreadsheet, having collected data from across multiple different EMRs, all in 1 place. [This] makes the data much more easily synthesized and analyzed.

What are the next steps for this line of research and further improving personalized cancer care?

Our study demonstrated that it was feasible; that we could consolidate all these data into 1 place. Now, the goal is to make sure that this is done, utilizing all the different databases across our health system, and increasing the number of variables that are available. As I mentioned, at the present time, we are learning what more is needed beyond the basics, such as cancer stage, disease sites, and treatments rendered. We are now learning what laboratory values [must] also be pulled in [and] what imaging data need to be pulled in. There are tons of variables that can be added to this registry to enrich it and make it as useful as possible across multiple domains.

Then, the next step is going to be for this to be adopted by other healthcare systems across the country. It has the potential to be a useful tool. Again, amongst the 3 domains that I mentioned—clinical, cancer registry, and research—another useful tool for this is [among] patients who are seeking to be enrolled onto clinical trials, often to determine whether or not a candidate is eligible for a clinical trial. A lot of information is needed about previous treatments that have been rendered, cancer stage, disease site, etc. Now, all of that is available in the database, so patients can be selected automatically for clinical trial eligibility. There are tons of ways that these data or this software can be utilized.

We are at the tip of the iceberg in terms of learning the various ways that we can use this software to enhance patient care and to personalize care for patients by determining what trials they might be eligible for and what the next best steps are in treatment based on their individual profile in the database.

Reference

Tchelebi L, Lindsay WD, Yee K, Wishinsky J, Labarca ME, Potters L. Building a comprehensive cancer database across fourteen treatment facilities in a large health care system. Presented at: 2025 American Society for Radiation Oncology Annual Meeting; September 27-October 1, 2025; San Francisco, CA.