Research & Publications｜Dr. Hwa-Yen Chiu｜Chest Medicine, Sleep Medicine and Medical Artificial Intelligence

Research & Publications

Hwa-Yen Chiu, MD

Chief of Pulmonary Medicine at Taipei Veterans General Hospital Hsinchu Branch. Institute of Biophotonics, National Yang Ming Chiao Tung University. His work spans artificial intelligence and radiomics in lung cancer imaging, sleep medicine, pulmonary function and small airway assessment, and respiratory critical care.

The purpose of this page is simple: so that patients, colleagues and search engines can all check what these judgements are built on.

🆔ORCID0000-0001-8759-7871 📚PubMedAuthor publication search 🏥Clinic informationTaipei Veterans General Hospital Hsinchu Branch

Research themes

🔬

Artificial intelligence and radiomics in lung cancer imaging

Deep learning for nodule detection on chest X-rays, and CT/MRI radiomics to predict response and progression-free survival on targeted and immune therapy. Alongside that, systematic reviews comparing how far AI really is from expert readers.

9 papers

😴

Sleep medicine

Phenotype work in obstructive sleep apnoea — the rapid-eye-movement-related type (REM-OSA), the features seen in younger people, and the link with nocturnal blood pressure change; plus interpretable machine learning to assess scoring consistency across several sleep centres.

4 papers

📈

Pulmonary function and small airway assessment

Using impulse oscillometry (IOS) to find small airway dysfunction in the group who have symptoms but normal lung function — people who in the past were often assumed to be fine because the blowing test came back normal.

1 paper

🫀

Respiratory critical care and clinical care

Prognostic factors in intensive care for patients with advanced lung cancer and decisions about do-not-intubate orders, ventilator-related complications, chemotherapy for neuroendocrine tumours, and a review of progress in COVID-19 treatment.

5 papers

Peer-reviewed publications

19 papers in all, ordered by year, newest first. The source is PubMed; click a PMID to see the original record and abstract.

2025Co-author

Effects of obstructive sleep apnea on nocturnal changes in blood pressure — a retrospective study.

Hypertension Research. 2025 Apr;48(4):1379-1388.

An analysis of 2,037 people who underwent polysomnography: sleep apnoea was independently associated with a morning rise in blood pressure, and most clearly so in the moderate-to-severe group.

PMID 39871002 DOI

2024Co-author

Personalized prediction of immunotherapy response in lung cancer patients using advanced radiomics and deep learning.

Cancer Imaging. 2024 Sep 30;24(1):129.

Radiomics of the tumour and of the vessels around it, combined with clinical features, in a DeepSurv model predicting progression-free survival after immunotherapy (AUC 0.76–0.80, C-index 0.83).

PMID 39350284 DOI

2024Co-author

Standalone deep learning versus experts for diagnosis lung cancer on chest computed tomography: a systematic review.

European Radiology. 2024 Nov;34(11):7397-7407.

A systematic review of 20 studies: deep learning reached 82% sensitivity and 75% specificity, against 81% and 69% for the experts. The difference in specificity was statistically significant, the difference in sensitivity was not, and performance varied with the imaging modality.

PMID 38777902 DOI

2024First author

Progress in Serial Imaging for Prognostic Stratification of Lung Cancer Patients Receiving Immunotherapy: A Systematic Review and Meta-Analysis.

Cancers (Basel). 2024 Jan 31;16(3):615.

A meta-analysis of 10 studies: delta radiomics predicted six-month treatment response with an AUC of 0.81, and the pooled hazard ratio for progression-free survival was 4.77.

PMID 38339369 DOI

2024Co-author

Radiomics of metastatic brain tumor as a predictive image biomarker of progression-free survival in patients with non-small-cell lung cancer with brain metastasis receiving tyrosine kinase inhibitors.

Translational Oncology. 2024 Jan;39:101826.

MRI radiomics of brain metastases used to build a CoxCC model predicting progression-free survival after EGFR-TKI treatment (c-index 0.75), which outperformed the existing prognostic indices for brain metastasis.

PMID 37984256 DOI

2023Co-author

Evaluating the Potential of Delta Radiomics for Assessing Tyrosine Kinase Inhibitor Treatment Response in Non-Small Cell Lung Cancer Patients.

Cancers (Basel). 2023 Oct 24;15(21):5125.

Serial CT analysis in 322 patients with advanced lung adenocarcinoma. Worth noting: predictive accuracy fell once data were pooled across centres, which shows that radiomics still needs standardisation.

PMID 37958300 DOI

2023Co-author

Undiagnosed dyspnea with pulmonary nodule.

Respirology Case Reports. 2023 Sep 27;11(10):e01226.

Case report: when airway disease is already under control, stay alert to other causes of breathlessness; the lateral chest X-ray is valuable for working out where a lung lesion sits in three dimensions.

PMID 37771849 DOI

2023Co-author

Application of Radiomics in Prognosing Lung Cancer Treated with Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitors: A Systematic Review and Meta-Analysis.

Cancers (Basel). 2023 Jul 8;15(14):3542.

A meta-analysis covering 3,111 patients, with a pooled hazard ratio of 2.80, a c-index of 0.71 and an AUC of 0.73; study quality was assessed with QUIPS and RQS.

PMID 37509204 DOI

2023Co-author

A radiomics-based deep learning approach to predict progression free-survival after tyrosine kinase inhibitor therapy in non-small cell lung cancer.

Cancer Imaging. 2023 Jan 20;23(1):9.

270 patients with stage IIIB–IV EGFR-mutant lung cancer, with 593 radiomic features extracted from pre-treatment CT to build a multiple-time-point DeepSurv prediction model.

PMID 36670497 DOI

2022First author

Obstructive sleep apnea in young Asian adults with sleep-related complaints.

Scientific Reports. 2022 Nov 29;12(1):20582.

A cohort analysis covering 2000–2017: young adults made up 25.1% of those having a sleep study and 19.8% of those diagnosed with OSA, and their symptoms were more marked and their hypoxia more severe. In young men, hypertension was associated with OSA risk.

PMID 36447027 DOI

2022First author

Artificial Intelligence for Early Detection of Chest Nodules in X-ray Images.

Biomedicines. 2022 Nov 7;10(11):2839.

U-Net to segment the lung fields, YOLOv4 to detect the nodules. Test sensitivity 79% (3.04 false positives per image). Looking back over 383 sets of chest X-rays from the five years before diagnosis, AI assistance extended the median time from detection to diagnosis from 8 days to 46 days — that is, it found them earlier.

PMID 36359360 DOI

2022First author

Clinical Characteristics of Rapid Eye Movement-Related Obstructive Sleep Apnea: An Experience in a Tertiary Medical Center of Taiwan.

Nature and Science of Sleep. 2022 Aug 31;14:1521-1532.

Of 1,331 patients with OSA, 31.1% had the REM-related type. After adjustment for severity, that type spent longer desaturated and reached a lower minimum oxygen saturation, and was associated with younger age, female sex and higher BMI.

PMID 36068886 DOI

2022Co-author

Prognostic factors for advanced lung cancer patients with do-not-intubate order in intensive care unit: a retrospective study.

BMC Pulmonary Medicine. 2022 Jun 24;22(1):245.

140 patients with advanced lung cancer who had a do-not-intubate order: those with a P/F ratio ≥ 150 and APACHE II < 16 had significantly better 28-day and 90-day survival, which can inform decisions in the intensive care unit.

PMID 35751074 DOI

2022First author

Tracheoesophageal fistula presenting with ventilator air leaks and recurrent pneumonia.

Respirology Case Reports. 2022 Mar 31;10(5):e0947.

Case report: a tracheo-oesophageal fistula after tracheostomy, presenting with ventilator air leaks and recurrent pneumonia, with the nasogastric tube visible at bronchoscopy.

PMID 35386577 DOI

2022First author

Application of Artificial Intelligence in Lung Cancer.

Cancers (Basel). 2022 Mar 8;14(6):1370.　(review article)

From image reading in the 1960s to FDA-cleared AI software, whole-slide image annotation, classification of EGFR mutation and PD-L1 expression, and on to prognosis and side-effect prediction — a full review of what AI is used for in lung cancer and of what stands in the way of putting it into practice.

PMID 35326521 DOI

2021Co-author

Large-scale assessment of consistency in sleep stage scoring rules among multiple sleep centers using an interpretable machine learning algorithm.

Journal of Clinical Sleep Medicine. 2021 Feb 1;17(2):159-166.

Six sleep centres in Taiwan, 679 patients: interpretable machine learning used to assess within-centre and between-centre consistency in sleep stage scoring, which successfully identified one centre with a quality problem.

PMID 32964831 DOI

2020Co-author

Overview of coronavirus disease 2019: Treatment updates and advances.

Journal of the Chinese Medical Association. 2020 Sep;83(9):805-808.　(review article)

A review of treatment progress in the early stage of the COVID-19 pandemic, covering prevention, diagnosis and the challenges of clinical management.

PMID 32520770 DOI

2020First author

Small Airway Dysfunction by Impulse Oscillometry in Symptomatic Patients with Preserved Pulmonary Function.

The Journal of Allergy and Clinical Immunology: In Practice. 2020 Jan;8(1):229-235.e3.

Of 349 symptomatic patients, 73.1% had normal lung function. The study set out IOS cut-offs for small airway dysfunction and showed that F_res is more sensitive than FEF_25-75. It is a direct answer to the clinical bind of “symptoms with normal lung function”.

PMID 31299351 DOI

2018First author

Toxicities, safeties and clinical response of dacarbazine-based chemotherapy on neuroendocrine tumors in Taiwan population.

Journal of the Chinese Medical Association. 2018 May;81(5):423-428.

An analysis of 48 Taiwanese patients with neuroendocrine tumours: the Ki-67 index predicted outcome in grade 3 cases and guided the choice of chemotherapy.

PMID 29287706 DOI

The list follows what PubMed indexes, so it may not cover Chinese-language journal articles, conference papers or book chapters that PubMed does not index.

How this research comes back to the clinic

The tools on this site were not made out of thin air. Several come straight from the work above:

The small airway work → An abnormal lung function report and the IOS reading in the smart report system （中文）: so that people whose lung function is normal but who still have symptoms know what else can be looked at.
The sleep apnoea work → the snoring self-check and the OSA health topic: including the REM-related type and the younger patients, the phenotypes that are easily missed.
The lung cancer AI and imaging work → the lung cancer screening assessment and the eligibility tool for the government-funded LDCT screening programme （中文）: and why the limits of AI have to be stated honestly.
The critical care and decision-making work → the respiratory care health topic: talking through do-not-intubate orders and care in the intensive care unit.

Academic exchange and collaboration

Enquiries about research collaboration, speaking invitations and data sharing are welcome, particularly on the clinical validation and real-world deployment of medical AI. If you are a patient, the appointment booking links are below.

AI Medical Lab Contact via LINE Book an appointment