Geomatic SAR Analytics
BY ALAIN CROTEAU AND KHALED BELHASSINE
© 2016 FrontLine Security (Vol 11, No 1)

The evaluation and integration of all SAR-related knowledge can help search managers consider and integrate many analytic factors that should help find missing wanderers in less time, and with better end results. An innovative and integrative geographic information system (GIS) tool has been developed by integrating the scientific knowledge of medicine, statistical data of Search and Rescue and the knowledge of SAR specialists, with the technology of Geomatics. This GIS tool effectively considers all different perspectives in five adaptable layers of information. Knowing that a SAR specialist can now rely on such a tool should bring relief to the many long-term caregivers and family members that have to deal with Alzheimer wanderers (AWs).

Development
SAR procedures for AWs have always posed a great challenge to SAR teams across Canada. The odd behavior of these patients makes it very difficult to predict their reaction, yet both family and care facility staff expect SAR specialists to perform independently of these challenges.

Information gathered from the lost person questionnaire, mixed with analyses from Koester statistics, can help prioritize a search area but it takes a great deal of experience to recognize and integrate all influential aspects while under the pressure of time and safety.

The research literature suggests there are four types of disorientation all linked to corresponding brain damage. Specifically, there is topographical or landmark agnosia (the inability to orient oneself in one’s surroundings as a result of focal brain damage); egocentric disorientation (unable to accurately reach for visual objects or state the relationship between two objects); heading disorientation (able to determine location using landmarks, but unable to determine which direction to proceed in order to reach their destination); and anterograde disorientation (marked by the inability to orient in new environment). Such disorientation can, too-often, lead to the person wandering away unnoticed.

While common SAR knowledge combines all types of dementia into the same profile, this new tool proposes two types of wandering – aimless or purposeful. We suggest these classifications will streamline future search tactics, making them more efficient.

Answers given by caregivers or family members to the profile questions above will establish the level of ‘purposeful’ wandering behaviour and should influence a search manager’s prioritization.

Narrowing the Search area
To help the Search Manager deal with the various types of information available, we developed a computer program that automatically generates a probability of area (POA) map. This tool is composed of an imagery map in which the roads are laid down and every intersection is mathematically identified. We call these intersections decision nodes.

The rationale for this tool is based on four elements:

  1. Despite the quality and quantity of resources deployed to a search area, AWs will never be found if they are not in the search area.
  2. All search areas are not of equal value. Even if a mathematical approach to aggregate opinions is used (such as the Mattson method) it can only equal the value of available and processed information.
  3. During a search, the amount and variety of information to evaluate and process is too important to be considered without tools.
  4. The behaviour of purposeful versus aimless wandering is too different to be considered one and the same.

Aimless wandering is a non-focused style of walk, with little to no apparent direction, purpose or destination. It may take place because the person is bored or feels the need to move about or may express the person’s response to feelings of stress, anxiety or physical discomfort.
Purposeful wandering is a goal-oriented type of walking where the AW wishes to accomplish something (such as go to the store for groceries). The person may appear to be searching for something or trying to return to familiar surroundings from their past. They may be looking for a familiar place that will give them feelings of security and reassurance or they may have a physical need, such as hunger or the need to use the washroom.

To incorporate all information and automatically generate a POA map, we developed five layers of information. Each layer gathers specific information that will influence the assigned weight of each decision node. All layers influence the subsequent layer to produce the fifth cumulative layer.

Layer 1: assigns a value to each decision node based on Koester statistical distance from the Initial Planning Point (IPP).

Layer 2: weighs the influence of elevation change. Statistics have shown that 81% of AWs are found on the same elevation or lower. However, our perception is that this statistic applies mostly to the aimless wanderer, therefore, our algorithm considers the profile determination result in its weight attribution.

Layer 3: considers the wanderer’s past history. We think this is especially true for purposefully-oriented AWs where the scene would serve as a stimulus. Therefore, it becomes important to gather, map, evaluate and consider the profile determination result in attributing a weight.

Layer 4: is associated with punctual investigative information. It brings to light a new concept we call ‘Point 100%’ or P100, and it can only be used if the time elapsed is less than 1 hour. The notion of P100 is relative to time and place. It is closely related to the last known position or to the place last seen. It too, needs to be corroborated to 100%; however, in contrast to other notions, it implies that the AW cannot be beyond a calculated distance based on travel speed.

A 2010 study on AWs, by G. Kemon, established their average speed at 2.9 kilometres per hour on flat terrain over 100 metres. The study focussed on a 100m stroll on flat terrain, however we felt that a slightly slower speed would more accurately reflect reality, so we used 2 km/h to calculate the decreasing weighted attribution rate.

If a direction of travel is established, a 70° angle is drawn (35° on each side of the travel direction). Koester’s statistics show that 95 percent of AWs are found within that angle. Again, the algorithm will consider all previous information. The result boosts up all decision nodes within the 70º angle while respecting the travel speed.

Layer 5: is the only POA map the user will see as it is the combined final product of the algorithm. It should allow the Search Manager to make an informed decision based on the available information. Using a color code to highlight the hot areas, this tool provides an easy and intuitive visual of the most promising search area.
 

Conclusion
In light of our findings, we ascertain that AWs seldom abruptly leave a path or road to enter a field or forest – they typically remain on a fairly straight path to their intended destination / goal (which is clear in their mind but the geography might be wrong). This finding is supported by the ISRID database that highlights 95% of AWs are found within 300 metres of any path or road, and that 95% are found in a 70° angle from their last direction of travel (Koester LPB). This finding is also supported by medical research illustrating the affects of Alzheimer’s disease on the frontal lobe (the area that plays a significant role in decision-making).

We also suggest that purposeful and aimless wanderers are influenced differently by elevation, affecting the aimless wanderer more. Could it simply be linked to the metabolic cost of going uphill without a clear reason to do so? Ongoing data collection should help us confirm our hypothesis.

The overall perspective of cumulative probability values allows the creation of an automated POA calculation tool. Our research clearly demonstrates that the use of such a tool provides a Search Manager with the ability to consider and integrate various factors that should help in finding missing AWs in less time – which usually means better end results.

The notion of a P100 illustrates the need for rapid communication, rapid analysis and rapid action. When possible, such information should result in a fast recovery.

A SAR Manager requires as much accurate information as possible to have any chance of success beyond pure luck. They need to gather, process and analyze a great deal of what may seem incompatible data. In many ways, the search function is a mathematical art. Technology can facilitate a team to search smarter and faster.

We hope to work with other specialist to simplify the tool and make it easily accessible in hopes of accelerating and helping Search Managers make life-saving choices.

Among ongoing developments, we hope to adapt the tool to a wilderness environment.

A tool that helps search managers successfully integrate a wide array of data for faster assessment will certainly bring relief to the many long-term caregivers and family members as they come to understand the grip the Alzheimers disease has on their loved one.

===
Alain Croteau is a SAR professional.
Khaled Belhassine is a researcher with Laval University.

NOTE: This POA software is working but ­currently needs a powerful computer to run it and advanced geomatic knowledge. The researchers are looking to team with an organization that can help bring this ­software to the next level of useability for SAR managers.

RELATED LINKS

Comments

CLICK HERE TO COMMENT ON THIS ARTICLE