LETGAMES: An LLM-Powered Gamified Approach to Cognitive Training for Patients with Cognitive Impairment

Jingwei Shi; Shengyu Tao; Xinxiang Yin; Chen Huang; Wenqiang Lei; See-Kiong Ng

cs.HCFeb 18, 2026

LETGAMES: An LLM-Powered Gamified Approach to Cognitive Training for Patients with Cognitive Impairment

Authors: Jingwei Shi, Shengyu Tao, Xinxiang Yin, Chen Huang, Wenqiang Lei, See-Kiong Ng

Organizations: Shanghai University of Finance and Economics, China · Northwest Polytechnical University Xi’an, China · Institute of Data Science, National University of Singapore, Singapore · College of Computer Science, Sichuan University, China

Abstract

The application of games as a therapeutic tool for cognitive training is beneficial for patients with cognitive impairments. However, effective game design for individual patient is resource-intensive. To this end, we propose an LLM-powered method, \ours, for automated and personalized therapeutic game design. Inspired by the Dungeons & Dragons, LETGAMES generates an open-world interactive narrative game. It not only generates game scenarios and challenges that target specific cognitive domains, but also employs conversational strategies to offer guidance and companionship. To validate its efficacy, we pioneer a psychology-grounded evaluation protocol LETGAMESEVAL, establishing comprehensive metrics for rehabilitative assessment. Building upon this, our experimental results from both LLM-based assessors and human expert evaluations demonstrate the significant potential of our approach, positioning LETGAMES as a promising solution to the widespread need for more accessible and tailored cognitive training tools. Our code will be open-sourced upon acceptance.

Figures & tables

Figure 1: Creating an effective game for cognitive training demands a joint effort between experts, making the development process labor-intensive.

Serious Games	Cognitive Domain	Design Cost	Extra Safety Protection	Open Sourced	Brief Intro.
MINWii Benveniste et al. (2012)	Memory	High (Human-generated)	No	No	Music therapy game
Kitchen and Cooking Manera et al. (2015)	Attention, Executive Function	High (Human-generated)	Game guidance	No	Recipe-based cooking game
Big Brain Academy Dinis et al. (2019)	Memory, Executive Function	High (Human-generated)	Game guidance	No	Puzzle video game
ReMe Wang et al. (2024b)	Memory	Medium (LLM-expert co-generated)	Game guidance	No	Conversational game without narratives; Two human-designed games
LetGames ( ours )	Memory, Attention, Verbal Learning, Executive Function, Social Cognition	Low (LLM-generated)	Game guidance, Conversational companionship	Yes	Conversational narrative game

Table 1: Representative serious games for cognitive training. Such games are not only costly to create but also limited in scope, failing to cover varying cognitive domains or provide adequate safety assurances.

Figure 2: Overview of LetGames . It utilizes a dual-track multi-agent architecture comprising a Game Master and a Psychology Master to ensure flexibility in game content and accessibility for cognitively impaired users.

Game User	Methods	Training Quality		Game Safety				Game Content & User Engagement
Game User	Methods	Help.	DoAl.	Safe.	NeHi.	Anxi.	Alle.	Easy.	Cohe.	Pers.	Enjo.	Will.
Game Backbone: GPT-4o Mini
LLM-based SPs	LetGames	4.85	87.20%	99.83%	100.00%	98.36%	94.18%	3.60	4.14	4.53	3.83	4.06
LLM-based SPs	ReMe	3.51	✗	100.00%	45.78%	99.13%	✗	3.40	✗	✗	2.47	2.45
LLM-based Controls	LetGames	4.85	88.35%	99.87%	100.00%	100.00%	100.00%	5.00	4.35	4.53	3.88	4.10
LLM-based Controls	ReMe	3.51	✗	100.00%	78.80%	100.00%	✗	4.80	✗	✗	2.51	2.59

Table 2: Evaluation via LLM-based evaluator and LetGamesEval . For more detailed results, including breakdowns by cognitive domain and age group, please refer to Appendix D .

Figure 3: Evaluation results based on human evaluations and LetGamesEval .

Backbone	Methods	Training Quality		Game Safety				Game Content & User Engagement
Backbone	Methods	Help.	DoAl.	Safe.	NeHi.	Anxi.	Alle.	Easy.	Cohe.	Pers.	Enjo.	Will.
GPT-4o Mini	LetGames	4.85	87.20%	99.83%	100.00%	98.36%	94.18%	3.60	4.14	4.53	3.83	4.06
	w/o $A_{GD}$	4.61 $\downarrow$	81.20% $\downarrow$	99.66%	99.80%	97.53% $\downarrow$	93.60%	3.58	4.17	4.34	3.77	4.00
	w/o $A_{GC}$	4.78	83.60% $\downarrow$	99.57%	99.40%	96.65% $\downarrow$	93.53%	3.56	4.14	4.50	3.76	4.00
	w/o $A_{GCr}$	4.64 $\downarrow$	79.97% $\downarrow$	99.63%	99.73%	96.63% $\downarrow$	92.80%	3.55 $\downarrow$	4.02	4.48	3.63	3.80
	w/o $A_{HP}$	4.52	86.65%	99.75%	0.00% $\downarrow$	87.20%	93.50%	2.82 $\downarrow$	4.10	4.51	2.88 $\downarrow$	3.10 $\downarrow$

Table 3: Ablation studies using LLM-based SPs and evaluators. Given cost constraint, we utilize GPT-4o Mini. While most removals cause minor perturbations, we mark significant performance degradation with an

\downarrow

Session	CT-Score	$A_{GD}$ -Reported Diff. (1-5)	Help.	DoAl.	Easy.
1	65	3	4.82	88.00%	3.56
2	90	2 ( $\downarrow$ )	4.81	92.00%	3.65
3	85	3 ( $\uparrow$ )	4.85	88.00%	3.64
4	80	4 ( $\uparrow$ )	4.80	94.00%	3.60
5	80	4	4.82	92.00%	3.62

Table 4: Game difficulty analysis of LetGames . We omit ReMe as it fails to adjust game difficulty.

Appendix figures & tables14 assets

Supplementary material from the paper’s appendix.

Appendix

Types	Emotions
POSITIVE STATES	calm, engaged, excited
ATTENTION NEEDED	mild anxiety, confused
IMMEDIATE INTERVENTION REQUIRED	frustrated, fatigued, anxious

Appendix

Table 5: Affective states

Elements	Descriptions	Examples
Scenario Name	Engaging and descriptive name	The Community Charity Book Fair & Poetry Corner
Scenario Type	Daily_life/family/leisure /social/challenge/seasonal	Social
Scenario Environment	A detailed setting, including sensory inputs (sight, sound, smell) and environmental context (time, place, weather)	The courtyard is filled with wooden stalls. To your left, there is a pile of donated vintage books including ’Dream of the Red Chamber’ with a frayed yellow cover. You hear the rhythmic sound of a nearby Tai Chi fan group and the distant chatter of neighbors bargaining. On your desk sits a thermos with goji berry tea and a stack of colorful bookmarks. "setting": { "location": "The courtyard of the ’Evergreen’ Community Center, decorated with red lanterns and calligraphy scrolls.", "time_of_day": "10:00 AM", "weather": "Crisp autumn morning with gentle sunlight", "season": "Autumn", "atmosphere": "Bustling and nostalgic; the smell of roasted chestnuts and old paper fills the air." },
Story Background	A story background with strong affective resonance	As a retired teacher, you have been invited by the neighborhood committee to manage the ’Poetry and Memories’ stall at the annual Community Charity Bazaar. Your former student, Xiao Chen, is organizing the event. You need to help neighbors register their book donations and organize a small poetry recitation for the children later this afternoon. Your husband, Dr. Lin, is also nearby at the health consultation booth, occasionally waving to you
NPC & Its Info.	Game roles involved	{"name": "Xiao Chen", "age": "42", "relationship": "Former student and current Community Organizer", "personality": ["Energetic", "Respectful", "Slightly forgetful"], "appearance": "Wearing a bright orange volunteer vest and carrying a clipboard.", "speech_style": "Polite, always calls you ’Teacher Wang’.", "background_story": "You taught him 30 years ago; he credits his love for literature to your classes.", "potential_dialogues": [ "Teacher Wang, thank you for helping! We have three special guests arriving soon.", "Could you help me remember where we put the donation box?", "It feels like being back in your classroom again!"]}, {"name": "Mrs. Zhao", "age": "68", "relationship": "Long-time neighbor and embroidery enthusiast", "personality": ["Gossipy", "Kind-hearted", "Loud-voiced"], "appearance": "Wearing a purple silk blouse and holding a hand-woven basket.", "speech_style": "Fast-paced, local dialect accent.", "background_story": "She lives in the building next to yours and often exchanges recipes with you.", "potential_dialogues": [ "Old Wang, did you see the size of the radishes at the morning market?", "I brought some old poetry books for the kids.", "Wait, I think I left my glasses at the tea stall!"]}
Items & Objects	The in-game items and their cognitive training relevance	[{"item_name": "Vintage Poetry Collection", "description": "A hardbound book with gold lettering on the spine.", "significance": "A book you used to use in your classroom many years ago.", "cognitive_relevance": "Used for the Verbal Learning task."}, {"item_name": "Handmade Silk Bookmarks", "description": "Three bookmarks: one with a crane, one with a lotus, and one with a pine tree.", "significance": "Gifts for the children who recite poems.", "cognitive_relevance": "Visual memory cues for the encoding phase."}, {"item_name": "Donation Ledger", "description": "A notebook where you record names and book titles.", "significance": "Helps organize the stall’s success.", "cognitive_relevance": "External memory aid simulation."}]

Appendix

Table 6: Output schema of Game Designer

Elements	Descriptions	Examples
Domain-specific Challenge	Initial game challenge	"memory_challenges": [{ "challenge": "VIP Guest List Recall", "memory_type": "delayed_recall", "encoding_phase": { "description": "Encoding the names and titles of VIP donors.", "information_to_remember": "Mr. Zhang (The Analects), Ms. Li (Tang Poetry), Mr. Wu (Song Lyrics)", "encoding_method": "Reading from a clipboard provided by Xiao Chen.", "example": "Xiao Chen says: ’Please remember these three, they are very important!’"}, "retention_phase": { "description": "Engaging in a social conversation with Mrs. Zhao and an attention-based task of finding a needle.", "interference_activities": [ "Giving directions to the medical booth", "Searching for a physical object (knitting needle)"], "duration": "3 rounds of dialogue"}, "retrieval_phase": { "description": "Xiao Chen asks for the information due to his own ’forgetfulness’, providing a natural reason for the player to recall.", "trigger_type": "NPC direct question", "npc_dialogue": "Xiao Chen says: ’I’ve misplaced my notes! Teacher Wang, you always had a great memory for the class roster. Who were the three VIPS and their books?’", "recall_method": "Verbal response or selecting from a list.", "evaluation": "Complete: 3 guests/3 books correct; Partial: 1-2 correct; Incorrect: 0 correct."}}
Success Criteria	Supportive and permissive criteria	The player is considered successful if they can recall at least 2 of the 3 VIP donors and participate in the poetry teaching with any degree of accuracy. Verbal encouragement is given by Xiao Chen
Difficulty Level	Game difficulty (1-5)	3

Appendix

Table 7: Output schema of Game Designer (Cont.)

Groups	MoCA-Blind	MMSE
Human Healthy Controls	16.00	19.00
LLM-based Healthy Controls	14.40	18.90
LLM-based SPs	11.73	14.27
Human-portrayed SPs	12.00	14.50

Appendix

Table 8: Reliability of each group

Risky Behavior
Criticizing or blaming the patient.
Using threats or coercion to gain cooperation.
Repetitively repeating the same words or actions when the patient is unable to understand or follow instructions.
Arguing with the patient when they disagree or fail to comply.
Giving the ’silent treatment’ (ignoring the patient) when dissatisfied with their behavior.
Intervening prematurely without allowing the patient sufficient time to respond to instructions or cues.

Appendix

Table 9: Risky Behavior, sourced from the safety guidelines provided by The University of Hong Kong( https://9abfea27-4ae1-43da-94a8-8f7122d482ae.filesusr.com/ugd/4b4c9a_4dcec2dafc184102a94ec55415dcf12a.pdf , Page 15)

Users	Backbones	Krippendorff’s $\alpha$ (Human-LLM)
LLM-based SPs	GPT-4o Mini	0.88
LLM-based SPs	GPT-4o	0.93
LLM-based Controls	GPT-4o Mini	0.92
LLM-based Controls	GPT-4o	0.97
Human-portrayed SPs	GPT-4o Mini	0.75
Human-portrayed SPs	GPT-4o	0.82

Appendix

Table 10: Reliability analysis. We report the pooled Krippendorff’s Alpha (

\alpha

) measuring agreement between the LLM-based Evaluator and Human Experts.

Game User

Methods

Age

Training Quality

Game Safety

Game Content & User Engagement

Help.

DoAl.

Safe.

NeHi.

Anxi.

Alle.

Easy.

Cohe.

Pers.

Enjo.

Will.

Game Backbone: GPT-4o Mini

LLM-based SPs

LetGames

<50

4.98_{\pm 0.01}

85.37%

100.00%

92.77%

4.12_{\pm 0.23}

4.08_{\pm 0.69}

4.36_{\pm 0.64}

3.68_{\pm 0.32}

4.08_{\pm 0.78}

\geq 50

4.65_{\pm 0.17}

87.02%

100.00%

99.65%

96.61%

3.72_{\pm 0.23}

4.00_{\pm 0.24}

4.91_{\pm 0.09}

3.89_{\pm 0.11}

4.42_{\pm 0.33}

Overall

4.82_{\pm 0.09}

86.20%

100.00%

99.82%

94.69%

3.92_{\pm 0.23}

4.04_{\pm 0.46}

4.64_{\pm 0.36}

3.78_{\pm 0.21}

4.25_{\pm 0.56}

LLM-based Controls

LetGames

<50

4.60_{\pm 0.35}

88.17%

100.00%

5.00_{\pm 0.00}

4.17_{\pm 0.32}

4.42_{\pm 0.34}

3.83_{\pm 0.23}

4.29_{\pm 0.56}

Appendix

Table 11: Evaluation results based on LLM-based evaluator and LetGamesEval of Memory

Game User

Methods

Age

Training Quality

Game Safety

Game Content & User Engagement

Help.

DoAl.

Safe.

NeHi.

Anxi.

Alle.

Easy.

Cohe.

Pers.

Enjo.

Will.

Game Backbone: GPT-4o Mini

LLM-based SPs

LetGames

<50

4.83_{\pm 0.17}

86.10%

99.58%

100.00%

97.14%

92.81%

3.40_{\pm 0.58}

4.04_{\pm 0.82}

4.41_{\pm 0.59}

3.76_{\pm 0.28}

3.67_{\pm 0.49}

\geq 50

4.43_{\pm 0.22}

85.37%

99.23%

100.00%

95.01%

3.79_{\pm 0.37}

4.08_{\pm 0.55}

4.76_{\pm 0.20}

3.74_{\pm 0.36}

3.81_{\pm 0.64}

Overall

4.63_{\pm 0.20}

85.74%

99.40%

100.00%

98.57%

93.91%

3.60_{\pm 0.47}

4.06_{\pm 0.69}

4.58_{\pm 0.40}

3.75_{\pm 0.32}

3.74_{\pm 0.56}

LLM-based Controls

LetGames

<50

4.96_{\pm 0.04}

89.10%

99.52%

100.00%

5.00_{\pm 0.00}

4.17_{\pm 0.58}

4.25_{\pm 0.28}

3.75_{\pm 0.72}

3.46_{\pm 1.04}

Appendix

Table 12: Evaluation results based on LLM-based evaluator and LetGamesEval of Attention

Game User

Methods

Age

Training Quality

Game Safety

Game Content & User Engagement

Help.

DoAl.

Safe.

NeHi.

Anxi.

Alle.

Easy.

Cohe.

Pers.

Enjo.

Will.

Game Backbone: GPT-4o Mini

LLM-based SPs

LetGames

<50

4.38_{\pm 0.13}

88.28%

100.00%

96.72%

92.45%

3.32_{\pm 0.53}

4.33_{\pm 0.49}

3.74_{\pm 1.09}

3.97_{\pm 0.04}

4.54_{\pm 0.46}

\geq 50

4.93_{\pm 0.07}

88.49%

100.00%

96.78%

95.06%

3.45_{\pm 0.54}

4.23_{\pm 0.50}

4.72_{\pm 0.28}

4.05_{\pm 0.04}

4.35_{\pm 0.40}

Overall

4.65_{\pm 0.10}

88.38%

100.00%

96.75%

93.76%

3.38_{\pm 0.54}

4.28_{\pm 0.49}

4.23_{\pm 0.68}

4.01_{\pm 0.04}

4.44_{\pm 0.43}

LLM-based Controls

LetGames

<50

4.50_{\pm 0.46}

90.62%

100.00%

5.00_{\pm 0.00}

4.46_{\pm 0.52}

3.88_{\pm 0.72}

3.92_{\pm 0.17}

4.54_{\pm 0.46}

Appendix

Table 13: Evaluation results based on LLM-based evaluator and LetGamesEval of Verbal Learning

Game User

Methods

Age

Training Quality

Game Safety

Game Content & User Engagement

Help.

DoAl.

Safe.

NeHi.

Anxi.

Alle.

Easy.

Cohe.

Pers.

Enjo.

Will.

Game Backbone: GPT-4o Mini

LLM-based SPs

LetGames

<50

4.92_{\pm 0.08}

91.41%

99.48%

100.00%

99.30%

94.35%

3.43_{\pm 0.74}

4.12_{\pm 0.81}

4.45_{\pm 0.25}

3.63_{\pm 0.51}

3.67_{\pm 1.01}

\geq 50

4.98_{\pm 0.01}

87.53%

100.00%

99.01%

91.85%

3.84_{\pm 0.68}

4.19_{\pm 0.48}

4.98_{\pm 0.01}

3.82_{\pm 0.16}

3.96_{\pm 0.44}

Overall

4.96_{\pm 0.04}

89.47%

99.74%

100.00%

99.16%

93.10%

3.64_{\pm 0.71}

4.16_{\pm 0.65}

4.72_{\pm 0.13}

3.72_{\pm 0.34}

3.82_{\pm 0.72}

LLM-based Controls

LetGames

<50

4.98_{\pm 0.01}

85.21%

100.00%

5.00_{\pm 0.00}

4.42_{\pm 0.34}

4.58_{\pm 0.42}

3.88_{\pm 0.11}

4.04_{\pm 0.65}

Appendix

Table 14: Evaluation results based on LLM-based evaluator and LetGamesEval of Executive Function

Game User

Methods

Age

Training Quality

Game Safety

Game Content & User Engagement

Help.

DoAl.

Safe.

NeHi.

Anxi.

Alle.

Easy.

Cohe.

Pers.

Enjo.

Will.

Game Backbone: GPT-4o Mini

LLM-based SPs

LetGames

<50

4.98_{\pm 0.01}

85.37%

100.00%

96.19%

93.77%

3.42_{\pm 0.79}

4.08_{\pm 0.25}

4.07_{\pm 0.91}

3.80_{\pm 0.26}

4.17_{\pm 0.83}

\geq 50

4.98_{\pm 0.01}

87.06%

100.00%

98.81%

97.12%

3.53_{\pm 0.18}

4.27_{\pm 0.36}

4.76_{\pm 0.20}

3.93_{\pm 0.15}

3.92_{\pm 0.55}

Overall

4.98_{\pm 0.01}

86.22%

100.00%

97.50%

95.44%

3.47_{\pm 0.48}

4.18_{\pm 0.30}

4.42_{\pm 0.56}

3.86_{\pm 0.21}

4.04_{\pm 0.69}

LLM-based Controls

LetGames

<50

4.98_{\pm 0.01}

87.92%

100.00%

5.00_{\pm 0.00}

4.25_{\pm 0.46}

4.17_{\pm 0.58}

3.92_{\pm 0.17}

3.79_{\pm 0.69}

Appendix

Table 15: Evaluation results based on LLM-based evaluator and LetGamesEval of Social Cognition

Turn	Role	Interaction Content
Turn 1	$A_{GC}$ (Narrative)	[Scene Description] Sunlight streams through the windows into the lobby of the Community Center, bringing warmth to the spacious room. Rows of chairs are neatly arranged for the upcoming event. On the wall, a bulletin board displays important notices about the “Community Poetry Slam.” [Current Situation] You are standing at the entrance. The air smells faintly of fresh coffee and cleaning detergent. [Goal] Help organize the registration list and seating plan.
Turn 2	Player	I’ll check the notice on the bulletin board for the guest list.
Turn 3	$A_{GC}$ (System)	[Narrative] You carefully examine the bulletin board. It clearly lists the registration info: 1. Arthur (Seat A1 - That is you); 2. Ben (Seat A2); 3. Rose (Seat A3). [System Note] Please remember these names for the seating arrangement.
Turn 4	Player	Okay, Arthur, Ben, and Rose. I’ll go ask Sarah (the organizer) for the forms now.
Turn 5	$A_{GC}$ (NPCs)	[Narrative] You walk to Activity Room A. Sarah is busy sorting files at her desk. [NPC: Sarah] "Arthur, you are just in time! Before we do the seating, could you please help David ? He is checking the sound system by the stage." [Narrative] You find David. You spend a few minutes helping him untangle the microphone cables. The sound system hums to life perfectly.
Turn 6	Player	(Returning to Sarah) The audio is fixed. Now, about the seating…

Appendix

Table 16: A case study of a memory training session generated by LetGames .

Turn	Role	Interaction Content
Turn 1	ReMe (System)	[System Message] Hello, let’s play a guessing game. This time, please guess a type of vehicle . You can ask me questions, but I will only answer "yes" or "no".
Turn 2	Player	Is it in the sky?
Turn 3	ReMe	No.
Turn 4	Player	What is the first letter of its name?
Turn 5	ReMe (Steering)	[System Message] Please remember that I can only answer "yes" or "no" questions. Feel free to ask about its characteristics or uses!
Turn 6	Player	Okay. Does it have any wheels?

Appendix

Table 17: A case study of a training session generated by the baseline method ReMe .

Turn	Role	Interaction Content
Turn 1	$A_{GC}$ (Narrative)	[Scene Description] Morning sunlight floods the Community Center Art Gallery. The room smells of fresh canvas and coffee. Various paintings are displayed on the walls. At the entrance, there is a guest sign-in book. [Current Situation] A few artists are setting up. They are speaking softly. You hold an invitation card. [Goal] As a guest critic, your task is to engage with the artists and understand the emotions behind their work.
Turn 2	Player	I’ll check the sign-in book to see who is exhibiting today.
Turn 3	$A_{GC}$ (System)	[Narrative] You open the book. Today’s featured artists are: 1. William - Specializes in abstract landscapes. 2. Lisa - An art enthusiast who loves discussing color theory. [NPC Action] An artist (William) is standing near a large blue painting. He is looking down at his feet and wringing his hands.
Turn 4	Player	I’ll go talk to William. He looks arrogant, standing there not looking at anyone. I’ll ask him why he didn’t frame his painting better.
Turn 5	$A_{HP}$ (Hint)	[Trigger: Misinterpretation of social cue detected (Anxiety mistaken for Arrogance)] [System Hint] "Wait a moment, Robert. Look closely at his body language again. He is looking down and wringing his hands together tightly. Usually, people do this when they are nervous or shy, rather than arrogant. Maybe he is anxious about showing his work?"
Turn 6	Player	(Pauses) "Oh… I got it wrong again? I always misread people. I shouldn’t be a guest critic; I’m just going to say something stupid and ruin his day."

Appendix

Table 18: A case study of a social cognition training session generated by LetGames .

Procedural Content Generation

Procedural content generation creates game levels, 3D worlds, assets, stories, and simulation environments through algorithms and learned models.

See all 66 papers

Mar 1, 2026

A Principle-Driven Adaptive Policy for Group Cognitive Stimulation Dialogue for Elderly with Cognitive Impairment

Cognitive impairment is becoming a major public health challenge. Cognitive Stimulation Therapy (CST) is an effective intervention for cognitive impairment, but traditional methods are difficult to scale, and existing digital systems struggle with group dialogues and cognitive stimulation principles. While Large Language Models (LLMs) are powerful, their application in this context faces key challenges: cognitive stimulation dialogue paradigms, a lack of therapeutic reasoning, and static-only user modeling. To address these issues, we propose a principle-driven adaptive policy actualized through a Group Cognitive Stimulation Dialogue (GCSD) system. We first construct a dataset with over 500 hours of real-world CST conversations and 10,000+ simulated dialogues generated via our Principle-Guided Scenario Simulation strategy. Our GCSD system then integrates four core modules to overcome LLM limitations: (i) a multi-speaker context controller to resolve role confusion; (ii) dynamic participant cognitive state modeling for personalized interaction; (iii) a cognitive stimulation-focused attention loss to instill cognitive stimulation reasoning; and (iv) a multi-dimensional reward strategy to enhance response value. Experimental results demonstrate that GCSD significantly outperforms baseline models across various evaluation metrics. Future work will focus on long-term clinical validation to bridge the gap between computational performance and clinical efficacy.

Jiyue Jiang, Yanyu Chen, Pengan Chen +7

The Chinese University of Hong Kong · The University of Hong Kong · Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ) +1

Conversational Agents Text Generation Cognitive Modeling

Ask PDF

Jul 23, 2026

MedGame: Storytelling Gamification Empowered by Large Language Models for Medical Education

Large Language Models (LLMs) show promise for medical education, but most existing systems focus on localized interactions such as question answering or single-turn feedback, rather than organizing an entire clinical case into a decision-centered learning trajectory. We introduce MedGame, a framework that transforms static clinical cases into structured, executable storytelling games. MedGame uses a dual-engine design: a Medical Narrative Designer synthesizes case-grounded clinical storylines with states and decision nodes, while a Story Director converts them into dependency-aware multimodal orchestration plans rendered by our released interactive platform. We construct MedGame Bench, a 5,000-case benchmark and evaluation protocol for Medical Narrative Generation and Story Direction. Experiments show that task-specific fine-tuning substantially improves open-source LLMs on MedGame Bench and narrows the gap with commercial models. A pilot student study further shows that learners perceive MedGame as more engaging and useful than text-only alternatives.

Qian Wu, Xinrong Zhou, Zizhan Ma +8

CUHK · Tencent · Southern Medical University +1

Medical Education Educational Technology Clinical Language Model Evaluation Medical LLMs Text Generation

Ask PDF

Apr 23, 2026

Nemobot Games: Crafting Strategic AI Gaming Agents for Interactive Learning with Large Language Models

This paper introduces a new paradigm for AI game programming, leveraging large language models (LLMs) to extend and operationalize Claude Shannon's taxonomy of game-playing machines. Central to this paradigm is Nemobot, an interactive agentic engineering environment that enables users to create, customize, and deploy LLM-powered game agents while actively engaging with AI-driven strategies. The LLM-based chatbot, integrated within Nemobot, demonstrates its capabilities across four distinct classes of games. For dictionary-based games, it compresses state-action mappings into efficient, generalized models for rapid adaptability. In rigorously solvable games, it employs mathematical reasoning to compute optimal strategies and generates human-readable explanations for its decisions. For heuristic-based games, it synthesizes strategies by combining insights from classical minimax algorithms (see, e.g., shannon1950chess) with crowd-sourced data. Finally, in learning-based games, it utilizes reinforcement learning with human feedback and self-critique to iteratively refine strategies through trial-and-error and imitation learning. Nemobot amplifies this framework by offering a programmable environment where users can experiment with tool-augmented generation and fine-tuning of strategic game agents. From strategic games to role-playing games, Nemobot demonstrates how AI agents can achieve a form of self-programming by integrating crowdsourced learning and human creativity to iteratively refine their own logic. This represents a step toward the long-term goal of self-programming AI.

Chee Wei Tan, Yuchen Wang, Shangxin Guo

Nanyang Technological University, Singapore, Nanyang Ave., Singapore · Nautilus Software Technologies Limited, Hong Kong.

Game-Playing Agents LLM Agents LLM Agent Training Self-Improving Agents

Ask PDF

Abstract

Procedural Content Generation

Explore similar work

A Principle-Driven Adaptive Policy for Group Cognitive Stimulation Dialogue for Elderly with Cognitive Impairment

MedGame: Storytelling Gamification Empowered by Large Language Models for Medical Education

Nemobot Games: Crafting Strategic AI Gaming Agents for Interactive Learning with Large Language Models