Step 4: Claims

Last modified by Vladimir Rullens on 2025/09/02 17:31

TopicQuestionAnswer

12.png

Measurements

For each positive and negative effect listed in step 3, describe how you could evaluate (measure) whether they actually occur.

Performance:

Robot: The AI is able to give required actions much faster than if a human was not present -> increase in manpower.

Patient: 

Caregiver: The caregiver can take of more tasks faster.

State:

Robot: 

Patient: Improved mood

Caregiver: Reduced stress and workload

Compare situation with and without AI, include statistics within the AI agent regarding how often it gets activated within a period of time and whether the action was effective in terms of mood and responsiveness and stress. Effectiveness may need to be handled by a third party.

For the caregiver give a questionnaire.

14.png

Benchmark

For each measurement, what are the benchmarks (criteria)? (i.e., what are desired values?)

Activations within a period of time: Its desired values should decrease over a period of time as the patient's situation improves. This ratio should be 1 or less.

Mood: emoticon_unhappy -> :I -> emoticon_smile (Score 0 to 1, increase mood by 10%). The time spent in a bad mood should be lower than without the AI, and their mood should actually increase.

Stress: Heartrate, ask them through a questionnaire (what is their mood like), or make a caregiver fill out the questionnaire for them. Questionnaire results should indicate less stress after the experiment.

Caregiver questionnaire: Questionnaire results should indicate reduced stress and experienced workload after the experiment.

 

13.png

Demonstration of AI-functionality

Can you describe how you could demonstrate that your AI function (s) achieve(s) the effects that you listed in the previous question?

As per what is described above, collect the indicated values and show a before and after. Then link that to our expected results.

I.e. The patient had 2-3 episodes a day that lasted for a couple of minutes. After including the AI in their day to day life this count reduced to 1-2. They were also handled much quicker by an average of 2 minutes.