This section defines the generic assembly of the model whereas the following section provide a case specific implementation and discussion.
3.3. Performance Measures
One of the most important issues to diagnose and treat ADHD patients is to determine a set of accurate performance measures. These performance measures should have the ability to differentiate accurately between children having ADHD symptoms and others who do not have it. According to psychiatric recommendations, it would be better to collect these performance measures from children within different environments such as at home and at school [
28].
There are many tests used to diagnose children with ADHD [
29]. Continuous Performance Tests (CPT) are the most popular laboratory-based test supporting clinical diagnosis [
30,
31,
32]. CPT is usually a computer-based test that aims to measure children attention and impulsivity. CPT involves the individual and random presentation of a series of visual or auditory stimuli that changes rapidly over a period of time. Children are informed to respond to the “target” stimulus and avoiding a “non-target” stimulus. The test provides summary statistics of performance parameters (e.g., response time, average response time, response time standard deviation, omission errors, and commission errors). These parameters have been shown to be useful in the detection of ADHD [
33]. An important limitation of traditional CPTs is low ecological validity [
34]. Ecological validity means the degree to which a performance test produces results similar to those produced in real life [
35]. One approach to improve assessment methods which offers better ecological validity is CPTs based on VR, such as the Aula Nesplora test [
36]. Those approaches have an advantage of being more realistic and ecologically-valid environment while still having the ability to assess the degree of ADHD severity. Using AR instead of VR will further improve the ecological validity of the performed test.
In this paper, multiple performance measures are used to provide ADHD diagnostics and treatment assessment. Thus, let us assume that we have an AR-game that frequently present an interactive environment with the following assumptions:
T: Number of tries in one session.
C: Number of correct tries in one session.
I: Number of incorrect tries (due to omission or commission errors) in one session.
K: Number of uncompleted tries in one session.
Then, the performance measures that will be used for providing ADHD diagnostics and treatment assessment include:
Correct Response Times (CRT): The percentage of measuring attention deficits for the time spent on the correct tries.
Mean of the CRT (
M): To compare with correct response time to make sure they follow opposite relation to one another.
Standard deviation of the
CRT (
SD): Indicative of impulsive and hyperactive symptoms.
Try time (): The maximum allowed time to complete one try within a session.
Omission errors (OE): The absence of any response during a try period to be used to measure inattention.
Commission errors (CE): The response to non-target stimuli which to be used to measure impulsivity.
Engagement Factor (
GF): It indicates the patient engagement level with the game.
Inattention Factor (
IAF): It represents the percentage of patient’s inattention.
Impulsivity Factor (
IMF): Indicative of percentage of the patient’s impulsivity observed in his/her behavior within a session.
Error Factor (
EF): It represents the percentage of the error rate during a session.
Correct Response Factor (
CRF): The percentage of the total correct response time relatively to maximum allowed time for all correct tries.
Performance Index (
PI): It reflects the single measure for the overall performance of the patient which depends on the correct response factor, error factor, and engagement factor.
Previous measurements are normalized within the interval (0–1). In Equation (3), CRT measures the length of time that the child takes to make a correct try (i.e., choose the target object in the try). The longer the CRT is, the more likely the child has attention deficit. This is because one of the symptoms of attention deficit is that the child cannot focus on tasks. As a result, the child takes a longer time compared to normal children when doing a task (i.e., choosing the target object in our case). We use the mean of all CRTs (M) in the game session to measure the attention deficit of the child. In addition, the standard deviation of CRTs (SD) is used in Equation (4) to indicate the impulsivity and hyperactivity of the child. The higher the value of SD, the more probability that the child suffers from impulsivity and hyperactivity. A child with impulsivity and hyperactivity has difficulty in controlling his moves after a certain period of time. As a result, the child starts periodically to move with no destination. Such a child in our case, will have great differences among CRTs because the impulsivity and hyperactivity will hinder him from moving towards the target object in some tries. Engagement Factor (GF), in Equation (5), indicates the engagement-level of the child in the game. In our case, the child is considered to be engaged in the game of s/he keeps playing the game. In contrast, the child is considered to be not engaged if s/he stops the game before completing all tries in the session. Thus, GF is defined as the number of correct and incorrect tries (C + I) divided by the total number of tries (T) in the session.
Inattention Factor (IAF) which is defined in Equation (6) indicates the child inattention. In our case, the child inattention increases when he makes Omission Errors (OE) in the session, i.e., when the patient does not choose any of the objects appearing to him/her. The number of uncompleted tries (K) in the session should be excluded when indicating IAF. Thus, IAF is defined as the number of OEs divided by . Impulsivity Factor () in Equations (7) is defined as the number commission errors divided by number of correct and incorrect tries. We also exclude K when defining IMF. In our case, the child who suffers from impulsivity will likely make more commission errors because impulsivity will prevent him/her from focusing when choosing an object. Error Factor (EF), given in Equation (8), indicates the child error rate. The error in our case includes omission and commission errors excluding . Thus, is equal to . Correct Response Factor (CRF), in Equation (9), measures the percentage of the correct response of the child in one session. In our case, CRF should be negatively affected by the amount of time that the child takes when he makes an incorrect try. Thus, we define CRF as the total summation of CRTs to the actual time of the game during the session (GT). In this case, CRF will be 100% if the child makes all tries correctly. Otherwise, it will decrease depending on the total amount of time spent by the child on incorrect tries.
The final performance measure which is given in Equation (8) is the Performance Index (PI). PI is a composite score which measures the overall performance of the child. In our case, we want the PI of the child to be affected positively by his CRF and negatively by his EF. In addition, we need to take into account different possible scenarios that can happen in the game session. One possible scenario is that the child does not complete all the tries in the session. The child can make one correct try and stop the game before finishing all the tries in the session. If we only considered the CRF and EF, the PI in the case would be the highest. In order to prevent this from happening, we consider the GF in the definition of PI. Another possible scenario is that we have two children who have the same CRF, EF, and GF but different GT. In this case, they will have the same PI. However, the child who has less GT should have a higher PI. Thus, we considered the ratio of GT to the maximum Session Time (ST) in the definition of PI. The PI of the child should be affected negatively by this ratio.