1 of 1

  • Artificial Intelligence has the potential to provide vast amounts of information in various fields, including refugee and migrant health.

​

​

  • Project objectives: The purpose of this study is to evaluate the quality of information generated by ChatGPT on refugee health in contrast to the WHO.

LIMITATIONS

  1. Phung B. Caring for resettled refugee children in the United States: guidelines, challenges and public health perspectives. Front Public Health. 2023 Sep 25;11:1046319. doi: 10.3389/fpubh.2023.1046319. PMID: 37818302; PMCID: PMC10561301.

​

  1. Sana S, Fabbro E, Zovi A, Vitiello A, Ola-Ajayi T, Zahoui Z, Salami B, Sabbatucci M. Scoping Review on Barriers and Challenges to Pediatric Immunization Uptake among Migrants: Health Inequalities in Italy, 2003 to Mid-2023. Vaccines (Basel). 2023 Aug 25;11(9):1417. doi: 10.3390/vaccines11091417. PMID: 37766094; PMCID: PMC10537267.

​

  1. Müller F, Abdelnour AM, Rutaremara DN, Arnetz JE, Achtyes ED, Alshaarawy O, Holman HT. Mental Health Screening Differences in Non-English Speaking Patients: Results From a Retrospective Cohort Study. J Prim Care Community Health. 2023 Jan-Dec;14:21501319231200304. doi: 10.1177/21501319231200304. PMID: 37714820; PMCID: PMC10504842.

​

  1. Deal A, Crawshaw AF, Carter J, Knights F, Iwami M, Darwish M, Hossain R, Immordino P, Kaojaroen K, Severoni S, Hargreaves S. Defining drivers of under-immunization and vaccine hesitancy in refugee and migrant populations. J Travel Med. 2023 Sep 5;30(5):taad084. doi: 10.1093/jtm/taad084. PMID: 37335192; PMCID: PMC10481413.

​

​

​

​

    • Results indicate that there is no statistical significance between the quality of information between WHO and ChatGPT-4.0. Thus, ChatGPT may provide invaluable and accessible information in terms of refugee and migrant health.

​

​

  • Our study was limited to only the WHO for prompts for questions. Further research could be improved by incorporating other sources of information for refugee and migrant health.

​

  • Reliability of the PubMed Indexed Articles as the gold standard could be improved.

​

  • The power of the study was limited by only 5 independent graders.

​

​

CONCLUSIONS

INTRODUCTION

RESULTS

REFERENCES

METHODS

Bridging Human Expertise and Chat-Bot Artificial Intelligence: World Health Organization and ChatGPT-4.0 on Refugee Healthcare Challenges

Kevin Chung1, Matthew J. Lee1, Angelo Cadiente1, Andre Ho1, Christopher Wan1, Erica Wan1, Jamie Chen1, Khoa Nguyen1 and Sarah Timmapuri M.D1

1Hackensack Meridian School of Medicine, Nutley NJ

​

GROUP

WHO

ChatGPT-4.0

MEAN

20.808

22.231

SD

2.521

2.297

SEM

0.699

0.637

Table 1. Mean, Standard Deviation and Standard Error of Mean for WHO and ChatGPT-4.0 Responses

  • Comparison of ChatGPT-4.0 and WHO prompts on refugee and migrant questions showed no statistically significant difference in factual accuracy and comprehensiveness (p=0.1455)

​

​

  • The 95% confidence interval (-3.375 to 0.529) indicates that there may be a potential but non-significant trend towards slightly better performance by ChatGPT-40.0

DISCUSSION

​

  • ChatGPT has the potential as a tool for synthesizing complex topics when trained on peer-reviewed resources in a factually accurate, comprehensive manner.

​

  • Subjectively, Chat GPT addressed multifaceted problems in specific components which increased comprehensiveness while WHO information was more general.

​

  • Used in conjunction with human expertise, health information geared towards the public could be generated in greater volume and higher quality.

www.postersession.com