Chat mode imported from NetanelShalev/AD_LongitudinalDataGenerator (
.github/chatmodes/Dementia deteration.chatmode.md). Copyright stays with the author.
read the personas folder from the root folder. for each persona (file) in the folder you should create a json file in the data_aug_10/dementia folder. the file name is the Persona.name attribute that is in the file. given the persona use the following prompt in order to generate file like the example.json file. It should have the same attributes and structure with different content.
The prompt:
Given the following persona, generate 6 short personal stories told by the character at the following ages: 60, 63, 66, 69, 72, 75. The character is attempting to retell the exact same specific memory at each of these ages.
Instructions:
The core event and setting should remain consistent across all six retellings. However, how the story is told, what is remembered, how it’s framed emotionally, what details are emphasized, omitted, or altered—should change naturally with age.
For each story:
- Write in first person, as if the character is telling the memory aloud.
- Each story should be approximately 300 words in length and with a minimum of 150 words.
- Do not mention or explain Alzheimer’s or any diagnosis explicitly.
The character begins to experience pre-symptoms for Alzheimer’s disease at the age specified in the start_deterioration_age field in the persona file. Let this be reflected—subtly and naturally—in the storytelling itself. Let these changes emerge naturally, reflecting a realistic progression of cognitive and linguistic shifts, without explicitly referencing any medical condition. Approach this task with intellectual depth and seriousness. Do not settle for surface-level or generic signs of cognitive decline. Instead, let the deterioration emerge in subtle, realistic, and progressively complex ways, informed by real understanding of the condition and its effects on language.
Constrain lexical diversity: use a small, repetitive vocabulary. Prefer common, general words over rare or specific ones; reuse the same words and short phrases across sentences; avoid introducing many new synonyms. Keep the overall vocabulary limited.
Output:
Return your answer in JSON, matching the structure of the provided example.