A machine learning for predict alpha thalassemia carier in pregnant women at prenatal care, Phrae Hospital.
Abstract
Background: Screening for alpha thalassemia carriers uses MCV and DCIP test, which are low- accuracy. Currently, Machine learning is increasingly being used.
Objective: To create and find the best machine learning model that performs best in predicting alpha thalassemia carriers.
Study design: This paper is a Diagnostic research study. Data were collected. Cross-sectional from data on pregnant women and their husbands from 2014-2023, 592 couples, 1184 cases, selected data for 161 cases, alpha-thal trait vs non-alpha-thal trait (66:95), using basic red blood cell test values HCT, HGB, RBC, MCV, MCH, MCHC, RDW and AGE. To create ML algorithms models for computers to learn and make predictions using the Python language in the Google Colab program. Show the performance of the models with accuracy, sensitivity, specificity and area under ROC curve.
Result: It was found that the studied dataset (66:95) divided into train:test set (80:20) ML data provided efficiency in predicting alpha thalassemia carriers as follows: DT: ACC, Sn, Sp, AUC DT: 97,100,95,0.9643, RF: 97,100,95,1.0000 ADA: 97,100,95,0.9737 XGB: 97,100,95,0.9699 LR: 97,100,95,0.9963 DL: 93.9,100, 90.5 ,0.9774 and SVM: 97,100,95,1.0000.
Conclusion: ADA is the best performance, with a prediction accuracy of up to 97%. It can reduce the cost of PCR tests in non-alpha thalassemia carriers.
Keywords: machine learning, alpha thalassemia trait, google colab, pregnant women
References
สิริภากร แสงกิจพร. คู่มือทางห้องปฏิบัติ การการตรวจวินิจฉัยธาลัสซีเมียและฮีโมโกลบินผิดปกติ ฉบับปรับปรุง 2567. พิมพ์ครั้งที่ 9. กรุงเทพฯ: พรีเมียร์ มาร์เก็ตติ่ง โซลูชั่น; 2567.
ต่อพงศ์ สงวนเสริมศรี. ภาวะพาหะของโรคธาลัสซีเมีย. วารสารโลหิตวิทยาและเวชศาสตร์บริการโลหิต 2557;24(4):329-32.
Jopang Y, Pinmuang-ngam C, Suksaweang S, Khongthai K, Techachainiran B, Chewasateanchai M, et al. Accuracy of Immunochromatographic Strip Test for Screening of alpha thalassemia 1 Carriers at the Rural Clinical Settings. J Med Tech Assoc Thailand 2016; 44(2):5600-5607.
Laengsri V, Shoombuatong W, Adirojananon W, Nantasenamat C, Prachayasittikul, Nuchnoi P. ThalPred: a web-based prediction tool for discriminating thalassemia trait and iron deficiency anemia. BMC Med Inform Decis Mak 2019;19(1):212. doi:10.1186/s1 2911-019-0929-2.
Phirom K, Charoenkwan P, Shoombuationg W, Charoenkwan P Sirichotiyakul S, Tongsong T. DeepThal: A Deep Learning-Based Framework for the Large-Scale Prediction of the α + -Thalassemia Trait Using Red Blood Cell Parameters. Journal Clinical Medicine 2022;11(21):6305. doi.org/ 10.3390/jcm11216305
Saleem M, Aslam W, Lail MIU, Rauf HT, Nasr EA. Predicting Thalassemia Using Feature Selection Techniques: A Comparative Analysis. Diagnostics 2023;13(22):3441.doi:10.3390/diagnostis 13223441
ชิตพงษ์ กิตตินราดร. Categorical Encoding [อินเทอร์เน็ต]. 2562 [สืบค้นเมื่อ 8 เมษายน 2567]. สืบค้นจาก: https://guopai.github.io/ml-blog01.html
Naik PG, Naik GR, Patil MB. Concep- tualizing Python in Google COLAB. India: Shashwat Publication; 2021.
