Winter 2017    |    www.tech-res.com
TRI - A Full Service CRO+

Saving Time in Clinical Operations through Data Visualization

PrecisionMedicineGraphic

Collection methods of raw research data continue to improve and capture more information than ever before. This creates several challenges for research organizations. Analyzing massive data sets without the proper tools takes time away from other research and regulatory compliance tasks. Time constraints become more problematic as demand increases for rigorous analysis of interactions within data sets.

 

Data visualization tools are used in a number of professional fields such as network engineering and finance to easily view how some data points relate to others. Did you know that these tools can also be used in clinical research? Data visualization tools can enable researchers to identify trends and observe interaction between specific factors within data sets. Data visualization tools provide insight that typical data management software cannot offer. Biomedical companies from around the world now utilize data visualization for projects ranging from clinical trials to international collaboration (such as AstraZeneca’s project to sequence 2 million genomes1).

 

Data Volume and Clinical Trial Operational Challenges


Modern drug development techniques use a diverse array of data points to guide Research and Development (R&D) decisions. As medical diagnostic technology continues to advance, new types of biomedical data will become available to researchers. This has driven the creation of methods and tools that integrate data (e.g. biomarkers) from all phases of research. These tools help identify potential risks to subjects taking an investigational product. While collected data has improved in quantity, quality, and diversity, researchers lack sufficient time to thoroughly review the data to better inform R&D activities.

 

Patient recruitment and retention is viewed as a major bottleneck for clinical trials. Nearly 80%2 of clinical trials fail to meet enrollment timelines. Enrollment challenges include3 short enrollment periods, insufficient numbers of staff available for recruiting patients, and disparate clinical research sites which are difficult to manage as a single unit. It is also difficult to efficiently identify and act on variables that influence recruitment. Particularly wide and long recruitment data sets can contain hundreds of factors that influence recruitment. Future study designs could be improved if researchers and CROs were equipped to analyze such factors and their interactions.

 

Post marketing safety monitoring is rapidly becoming more unwieldy as new drugs enter the market place. In particular, adverse reaction data analysis of patients taking multiple drugs is becoming more burdensome. While tools exist to help monitors obtain drug information, they come with major limitations. The most limiting issues with current4 tools are a lack of simultaneous multi-drug review features and overly complex presentation of information found. Additional tools to parse through this complex information only present the most dangerous and common adverse reactions. Thus if sponsors and regulatory monitors wish to obtain a complete picture of drug interactions, they must do so through troublesome methods. They may quickly find themselves overwhelmed with the sheer amount data they must compare. This increases time required to thoroughly review all relevant information.

 

Saving Time Through Data Visualization


These activities share a common problem; large amounts of data to review in a limited timeframe. Effectively managing a clinical trial becomes difficult when time must be taken from other tasks to fulfill data analysis requirements. Furthermore, data sets that are not thoroughly analyzed because of lack of time or efficient methods can lessen a trial’s impact and potentially put subjects at risk. In order to address these time constraints, a number of contract and biomedical research organizations have used data visualization to successfully save time and bring added value to their research efforts.

 

Drug Development Safety Monitoring Analytics


https://d1bp1ynq8xms31.cloudfront.net/wp-content/uploads/2013/06/NKI1.jpg

SOURCE: https://d1bp1ynq8xms31.cloudfront.net/wp-content/uploads/2013/06/NKI1.jpg

 

Researchers have claimed5 that data visualization is imperative in all clinical development phases for exploring and integrating biomarkers for multiple compounds side-by-side for decision making. This process uses input and collaboration between multidisciplinary experts. A data visualization system streamlines data review by enabling easier collaboration and reconciling interpretation of data (e.g. PK/PD) obtained during Phase I-IV. A number of studies5 have demonstrated the benefit of combining data visualization systems in various phases of pre-marketing trials.

 

Ayasdi is an example of a clinical trial and drug development data visualization tool. This system combines the utility of numerous analytic systems with machine learning. Researchers have used the system to plan for patient interventions ahead of their occurrence. An example6 of this can be seen in research conducted by the Michael J. Fox Parkinson’s Foundation. The Foundation used Ayasdi to track patient treatment outcomes between experimental and control groups more effectively than with previous tracking methods. The system also helped researchers find two previously unidentified unique sub groups within the patient pool.

 

Recruitment and Retention Analytics


http://blogs.sas.com/content/jmp/wp-content/blogs.dir/3/files/2015/07/blog11.png

SOURCE: http://blogs.sas.com/content/jmp/wp-content/blogs.dir/3/files/2015/07/blog11.png

 

Recruitment and retention data points are anticipated to grow more complicated as multi-site clinical trials7 become more common place. To address these complications, data visualization tools have been used to present accrual data in an easily understood format. Dashboard data visualization can enable informed decision making based on variables of interest. For instance, a custom dashboard system for subject recruitment was tested in two observational studies at the University of Louisville. Investigators were able to monitor the process of patient enrollment throughout the day in a variety of visual formats custom tailored for the study3. At the study’s conclusion, 77% of the Investigators using the system to obtain enrollment information found it easier or much easier to use as a means of communication compared to direct communication.

 

Tibco Spotfire is a data visualization tool that can monitor patient recruitment rates and identify data point interactions within recruitment data. A number of clinical trials under the National Cancer Institute (NCI) have their patient recruitment levels monitored and analyzed by a customized Spotfire data visualization system developed by Technical Resources International (TRI). These analytic systems allow TRI to provide NCI with actionable information for a number of clinical research aspects. Their data visualization solution can identify recruitment risks and opportunities, compare recruitment impact between potential study sites, and provide insight that informs future study designs.

 

The utility of predictive analytics can also be harnessed to augment a trial and increase its operational efficiency. JMP, a predictive analytics tool from SAS, can help researchers better foresee recruitment process threats and to take action accordingly. One of its functions allows users to assess the current and future state of recruitment at different study sites. This information is plotted8 out in a graph to provide researchers with a comprehensive overview of their study’s recruitment prospects.

 

Post Marketing Safety Monitoring Analytics



SOURCE: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3675775/figure/F3/

 

Cross referencing drug information for post marketing trials is difficult due to ineffective data systems. Fortunately, there are a number of data visualization tools that can enable robust comparative drug interaction analysis. For example, decision support systems with integrated data visualization were assessed for their impact on hastening potential adverse drug event reviews. The system tested included a database with 16,340 unique drug and side-effect pairs that represented 250 common medications. The system reduced medication side-effect research time by 60%9 when compared to traditional drug information sources.

 

Data mining algorithms and data visualization combined10 with electronic health care data (HCD) sets have also been used to generate informative representations of data from patients taking a variety of drugs. HCD is unique in that it provides temporal patient information. This inherent characteristic allows drug safety analysis to be visualized and analyzed using patient time graphs. This enables researchers to further analyze post-market data and possibly predict patient specific hazard periods.

 




Footnotes

  1. Ledford, H. (2016). AstraZeneca launches project to sequence 2 million genomes. Retrieved November 13, 2016, from http://www.nature.com/news/astrazeneca-launches-project-to-sequence-2-million-genomes-1.19797
  2. Haas, J. (n.d.). Web-Based Patient Recruitment. Retrieved November 13, 2016, from https://www.cuttingedgeinfo.com/preview/web-based-patient-recruitment-wp211/
  3. Mattingly, W. et al. (2015). Real-time enrollment dashboard for multisite clinical trials. Retrieved November 15, 2016, fromhttp://www.sciencedirect.com/science/article/pii/S2451865415300119
  4. Lamy, JB., Venot, A., Bar-Hen, A., Ouvrard, P., and Duclos, C. (2008). Design of a graphical and interactive interface for facilitating access to drug contraindications, cautions for use, interactions and adverse effects. Retrieved November 15, 2016, from http://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/1472-6947-8-21
  5. Brynne, L., Bresell, A., and Sjogren, N. (2013). Effective visualization of integrated knowledge and data to enable informed decisions in drug development and translational medicine. Retrieved November 15, 2016, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3842641/
  6. Alsumidaie, M. (2014). How Data is Transforming Clinical Trials and Healthcare: An Interview with Ayasdi's Pek Lum. Retrieved November 15, 2016, from http://www.appliedclinicaltrialsonline.com/how-data-transforming-clinical-trials-and-healthcare-interview-ayasdis-pek-lum
  7. Kernan, W. et al. (2011). Boosting enrolment in clinical trials: validation of a regional network model. Retrieved November 15, 2016, from https://www.ncbi.nlm.nih.gov/pubmed/21824978
  8. Jiang, P. (2015). Are we on schedule? Predictive modeling for patient recruitment in multicenter trials. Retrieved November 15, 2016, from http://blogs.sas.com/content/jmp/2015/08/03/are-we-on-schedule-predictive-modeling-for-patient-recruitment-in-multicenter-trials/
  9. Duke, J., Li, X., and Grannis, S. (2010). Data visualization speeds review of potential adverse drug events in patients on multiple medications. Retrieved November 15, 2016, fromhttp://www.sciencedirect.com/science/article/pii/S1532046409001567
  10. Harpaz, R. et al. (2012). Novel Data Mining Methodologies for Adverse Drug Event Discovery and Analysis. Retrieved November 15, 2016, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3675775/




About the Author

 

Victor Poonai is involved in safety committee organization, process development, and accrual data validation. He has been with TRI for nearly five years and has experience with subject matter including biodefense, parasitology, and human enhancement. Please visit www.tech-res.com for more information about TRI's services


TRI - Challenge Convention. Exceed Expectations. © Technical Resources International, Inc.  •  info@tech-res.com  •  Phone: 301-564-6400