The initial goals of our collaborative group was to determine how to share information that is typically not shared and to develop or find a platform in which to do that. Much of the information that is collected by government, non-profit, and educational organizations is considered proprietary and is held for organizational use only. The proprietary nature of data often makes understanding a subject that is important to the community across a number of sectors fragmented and segmented.
In creating this cooperative venture, the members of the group wanted to answer several questions:
- Was it possible to share proprietary data in a way that was acceptable to the owner of the data?
- Could we share data in a way that protected individual privacy and confidentiality?
- Could we share data in a platform that restricted access to only those who needed access?
- Would the interpretation and analysis of data from different sources with different centers of focus lead to a better understanding of the subject?
- Do the results provide information useful to those who supplied the data?
- Can benefit of the information be extended beyond the original scope of the project?
The maps that are depicted below show that we were successfully able to (1) develop a plan to share sensitive data that protected the privacy and confidentiality of the information in a way that was acceptable to owners of the original point-based data, and (2) use a data sharing platform that allowed for restricted access to only those who needed to manipulate and evaluate the data. The question is whether the two data sets produce information and learning that can deepen our understanding of the topic more than viewing one of the data sets independently.
Figure 1 presents the police service call data from the city of Allentown in a hot spot analysis map. From the analysis of point data over a 5-year period from 2011-2015, we were able to determine that parts of the city had a greater density of mental health calls than other parts of the city. The so-called “hot spots”, or areas of statistically significant clustering, are depicted in red.
It is important to note that this is incident-based data, which means the hot-spot areas identified in this map are those areas where the police were actually dispatched to the scene.
The object of our data sharing exercise was not to use single-source data when trying to develop a comprehensive understanding of our topic of interest. It was, and is, the belief of our group that in order to truly understand any subject we need to collect and analyze data from multiple sources. The more diverse that data which we attempt to collect, the better our overall understanding of the subject will be. To highlight this proposition, Figure 2 displays the same type of hot-spot analysis performed on emergency room and inpatient admissions supplied by the Lehigh Valley Health Network over the same 5 year period. This data set is important because its central focus is different. Rather than focusing on the location of the incident, as is the case with police data, hospital data reflects where the patient lives.
Interestingly, the data show a strikingly similar pattern when compared to the police service call data. (map) The data suggest that those who visit the emergency room at one of LVHN’s facilities live in the areas where police respond most often to mental health related calls-for-service.
This data overlay can inform both sectors in developing training and services.
If, as an example, the department was to consider some type of crisis intervention training it would be beneficial to provide training to the officers that most often patrol the areas shown on this map. From a healthcare perspective, it can help identify where preventative services can be deployed.
The final question to be answered is probably best answered by the members of the group who are gathered for this presentation. Does the data have value to groups other than the police and hospital. If so, who might benefit from the sharing of this sample of data and what other data might be relevant if it were shared?
- School attendance, performance, and transiency
- Social service delivery
- Community and economic development