Search

The concept of Big Data is changing the way that clinical research can be performed. Cardiothoracic surgeons need to understand the dynamic digital transformation taking place in the healthcare industry. In the last decade, technological advances and Big Data analytics have become powerful tools for businesses. In healthcare, rapid expansion of Big Data infrastructure has occurred in parallel with attempts to reduce cost and improve outcomes. Many hospitals around the country are augmenting traditional relational databases with Big Data infrastructure. Advanced data capture and categorization tools such as natural language processing and optical character recognition are being developed for clinical and research use, while Internet of Things in the form of wearable technology serves as an additional source of data usable for research. As cardiothoracic surgeons seek ways to innovate, novel approaches to data acquisition and analysis enable a more rigorous level of investigatory efforts.

In the first part of this series, we introduced the tools of Big Data, including Not Only Standard Query Language data warehouse, natural language processing (NLP), optical character recognition (OCR), and Internet of Things (IoT). There are nuances to the utilization of these analytics tools, which must be well understood by clinicians seeking to take advantage of these innovative research strategies. One must recognize technical challenges to NLP, such as unintended search outcomes and variability in the expression of human written texts. Other caveats include dealing written texts in image formats, which may ultimately be handled with transformation to text format by OCR, though this technology is still under development. IoT is beginning to be used in cardiac monitoring, medication adherence alerts, lifestyle monitoring, and saving traditional labs from equipment failure catastrophes. These technologies will become more prevalent in the future research landscape, and cardiothoracic surgeons should understand the advantages of these technologies to propel our research to the next level. Experience and understanding of technology are needed in building a robust NLP search result, and effective communication with the data management team is a crucial step in successful utilization of these technologies. In this second installment of the series, we provide examples of published investigations utilizing the advanced analytic tools introduced in Part I. We will explain our processes in developing the research question, barriers to achieving the research goals using traditional research methods, tools used to overcome the barriers, and the research findings.

Objective: Though interest in expansion of the use of less-invasive therapies among operable non-small-cell lung cancer (NSCLC) patients is growing, it is not clear that post-treatment surveillance has been comparable between treatment modalities. We sought to characterize institutional surveillance patterns after NSCLC therapy with stereotactic body radiation therapy (SBRT) and lobectomy., Methods: NSCLC patients treated with lobectomy or SBRT (2005 to 2016) at a single institution were identified. Natural language processing searched data fields within axial surveillance imaging reports for findings suggestive of recurrence. Duration and patterns of institutional surveillance were compared between the 2 groups., Results: Three thousand forty-two patients (73.5% lobectomy, 26.5% SBRT) met inclusion criteria. Patients had a longer median duration of surveillance after lobectomy (28.0 months vs SBRT 12.3 months, P < 0.001) and were more likely to undergo histopathological evaluation of clinically suspected relapse (206/274 [75.2%] vs SBRT 54/113 [47.8%], P < 0.001). Patients with clinical suspicion of recurrence had longer durations of institutional surveillance than those who did not among both cohorts (lobectomy 44.4 months vs 25.9, P < 0.001; SBRT 27.9 vs 10.3, P < 0.001). Landmark analyses at 1 and 3 years after therapy identified associations between receipt of lobectomy and ongoing surveillance at each time point (1 year odds ratio [OR] 2.10, P < 0.001; 3 years OR 1.71, P < 0.001) among all patients and those with documented stage I disease., Conclusions: We identified potential heterogeneity in institutional surveillance patterns after treatment of NSCLC with 2 therapeutic modalities. As less-invasive treatment options for operable patients expand, it will be critical to implement rigorous surveillance paradigms across all modalities.