Applying big data and artificial intelligence to melanoma management and research
10.3760/cma.j.cn114453-20210530-00238
- VernacularTitle:黑色素瘤大数据人工智能科研平台的建设及应用
- Author:
Jia FENG
1
;
Zhiwei CHEN
;
Jiaqi LIU
;
Jianying GU
Author Information
1. 复旦大学附属中山医院整形外科，上海 200032
- Keywords:
Melanoma;
Big data;
Artificial intelligence;
Machine learning;
Medical informatics applications
- From:
Chinese Journal of Plastic Surgery
2022;38(11):1222-1228
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To present a melanoma research platform, a big data and artificial intelligence tool, to realize the structured storage and accurate management of medical data, and improve the efficiency of clinical research.Methods:The multi-system data of the period from October 2007 to September 2020 was extracted from Zhongshan Hospital, Fudan University. The data was standardized, structured, and fully integrated. The big data technology was used to extract and integrate the data from various subsystems in the hospital, and a melanoma big database was established. Through artificial intelligence algorithm the disease characteristics were deeply mined, and the survival analysis was carried out using in-database algorithms.Results:The big data and artificial intelligence research platform for melanoma provides data screening and export, statistical analysis, original record tracing, and automatic selection of analysis algorithm. The melanoma platform achieved the aims of data visualization, trials preview, second response, and artificial intelligence automatic data extraction and storage. It can build a closed-loop research procedure efficiently. After screening, 152 patients with complete information were collected. There were 78 males and 74 females; the median age was 61 years (range, 30-97 years). Foot (133 patients) was the most common site of tumor, while groin (2 patients) and back (2 patients) were the least common site of tumor. The maximum diameter of the tumor ranged from 1 to 80 mm, with an average of 23.55 mm and a median of 20 mm. The most common Clark’s level was Ⅳ (62 patients). Ulceration was found in 17 patients. 10 patients had lymph node metastasis. Lung (5 patients) was the most common distant metastasis organ. More than 80% of patients’ Breslow thickness was less than 1 mm. Survival analysis showed that age ≥ 70 years, Clark’s level ≥ Ⅳ and Breslow thickness ≥ 4 mm were risk factors for poor prognosis ( P<0.05). Conclusions:The big data and artificial intelligence research platform for melanoma realizes the structured storage and accurate management of clinical data regarding melanoma, which provides a promising solution for the efficient use of clinical data.
