Experience and future perspective of artificial intelligence (AI) in postoperative gastric cancer treatment recommendation: a retrospective, observational, cross-validation study with multidisciplinary team board

doi:10.21203/rs.3.rs-2469053/v1

Download PDF

Research Article

Experience and future perspective of artificial intelligence (AI) in postoperative gastric cancer treatment recommendation: a retrospective, observational, cross-validation study with multidisciplinary team board

https://doi.org/10.21203/rs.3.rs-2469053/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Recently, owing to significant growth in the amount of information produced by cancer research, staying abreast of the developments has become a challenging task. Artificial intelligence (AI) can learn, reason, and understand the enormous corpus of literature available to the scientific community. However, large-scale cross-validation studies comparing the recommendations of AI and multidisciplinary tumor boards (MTB) in gastric cancer treatment have rarely been performed. Therefore, we retrospectively conducted a real-world study to assess the level of concordance between AI and MTB treatment recommendations.

Methods: We retrospectively analyzed the treatment recommendations of Watson for Oncology (WFO) and MTB for 322 patients with gastric cancer from January 2015 to December 2018 and compared the degree of agreement between them. The patients were divided into concordance and non-concordance groups. The factors affecting the concordance rate were analyzed.
Results: The concordance rate between AI and MTB was 86.96% at consideration level (280/322). The concordance rate for stage I gastric cancer was the highest (96.93 %). The concordance rates for stages II and III were 88.89% and 90.91%, respectively, which were close to 90%; however, the concordance rate for stage IV was the lowest at 45.83%. In the multivariate analysis, age, performance status, and stage IV gastric cancer had a significant effect on concordance between MTB and WFO.
Conclusions: The factors affecting the concordance rate were age, performance status, and stage IV gastric cancer. For increasing the validity of future medical AI systems for gastric cancer treatment, their supplementation of the local guidelines and the ability to comprehensively understand individual patients is essential.

Gastric cancer

Artificial intelligence

Multidisciplinary tumor board

Watson for Oncology

Concordance rate

While the amount of medical literature that needs to be covered is increasing rapidly every day, the amount of time available with medical specialists is limited. To account for this limitation, the medical field is subdivided and specialized, and a multidisciplinary team board (MTB) has been introduced for smooth communication between specialists in each subdivided medical field. MTB is a physical or virtual meeting in which different specialists converge to discuss clinical cases and determine the best diagnostic or therapeutic method. MTB is recommended in a variety of cancer guidelines and is the principal approach for managing patients with cancer in many countries (1).

However, owing to the incredibly rapid expansion of medical literature and advances in techniques, staying aware of these updates is challenging even for specialists in subdivided medical fields (2). Thus, it will become more difficult for MTB specialists to present individualized treatment recommendations for cancer patients based on the latest medical research. In this situation, the application of artificial intelligence (AI) in clinical practice could be a good solution for assisting MTB in managing patients with cancer.

AI can support clinicians in the diagnosis, treatment, and prognosis of a variety of diseases by the quick analysis of complex medical data. In addition, AI tools in medical imaging can be used to analyze images, such as X-ray, CT, and MRI images, and other findings that human radiologists might miss. Therefore, AI has the potential to compensate for human shortcomings and provide better treatment recommendations (3, 4). Therefore, reliance on AI in the management of cancer patients is expected to increase over time. However, for AI to be applied in actual medical practice, it must be validated using a pre-existing principal method.

To investigate the validity of AI in managing cancer patients, previous studies have compared the treatment recommendations of MTB and AI and analyzed the concordance rate. AI has shown a high concordance rate of 80–90% with MTB in breast and colorectal cancer. In contrast, the concordance rate of AI for gastric cancer varies by country (5). In addition, unlike other cancers, there are only few studies on AI for gastric cancer, and they either included only advanced stages of gastric cancer or the sample size for each stage was very small (6–8).

Therefore, we performed a large-scale cross-validation study to investigate factors that lower the concordance rate between medical AI and MTB in the recommendation of postoperative gastric cancer treatment. If these factors are supplemented, it is expected that AI will be able to better assist MTB as one of the members in actual clinical practice.

Patients and study design

We studied the treatment recommendations for consecutive 322 gastric cancer cases and compared the degree of agreement between Watson for Oncology (WFO) (IBM Watson Health, Cambridge, MA) and a 7-member MTB comprising two surgeons, two medical oncologists, a pathologist, a radiologic oncologist, and a radiologist at the Daegu Catholic University Medical Center (DCUMC), Daegu, Korea, from January 2015 to December 2018. Patients who fulfilled the following criteria were eligible for this study: (1) diagnosed with gastric cancer and the results confirmed using gastroduodenoscopy and histologic examination of resected specimens, and (2) surveillance or adjuvant chemotherapy after curative gastrectomy, or palliative chemotherapy in stage IV. The exclusion criteria were as follows: (1) neoadjuvant chemotherapy or conversion surgery; (2) history of other malignancies within 5 years; (3) incomplete data; and (4) younger than 18 years of age. Gastric cancer stage was determined according to the 7th edition of the American Joint Commission on Cancer (AJCC). The clinicopathological characteristics of the patients are summarized in Table 1. Patients with stage IV disease progression following systemic therapy (second-line and beyond) were also excluded. The study design was reviewed and approved by the Institutional Review Board (IRB) of DCUMC (DCMC-CR-21-017).

Table 1

Baseline characteristics of patients.
Variables		(n = 322) n(%) or mean(SD)
Age	Total	64.43(12.42)
	< 80	285(88.5)
	≥ 80	37(11.5)
Sex	Female	95(29.5)
	Male	227(70.5)
Performance status	0	227(70.5)
	1	44(13.7)
	2	26(8.1)
	3	25(7.8)
Tumor location	Proximal	80(24.8)
	Middle	80(24.8)
	Distal	162(50.3)
HER2 status	Negative	291(90.4)
	Positive	31(9.6)
Lympho-vascular invasion	No	200(69)
	Yes	90(31)
	Not available†	32
Perineural invasion	No	200(69)
	Yes	90(31)
	Not available†	32
Serum bilirubin	Total	0.57(0.49)
	< 1.2	306(95.0)
	≥ 1.2	16(5.0)
Serum Creatinine	Total	0.75(0.23)
	< 0.9	238(73.9)
	≥ 0.9	84(26.1)
Concordance	No	42(13.0)
	Yes	280(87.0)
Cancer stage	Ⅰ	163(50.6)
	Ⅱ	45(14)
	Ⅲ	66(20.5)
	Ⅳ	48(14.9)
Abbreviation; SD: Standard deviation; HER2: Human epidermal growth receptor 2
Not available†: Data from patients with palliative chemotherapy

Ai And Mtb Concordance Determination

We used the WFO version 16.4 as a medical AI. WFO is a well-known decision support system that can learn, reason, and understand the enormous corpus of literature available to the scientific community. In gastric cancer treatment, WFO considers not only disease stage, postoperative pathologic findings, human epidermal growth factor receptor 2 (HER2) status, and general condition (critical disease and performance status), but also the age at diagnosis, sex, weight, histologic type, and prior therapy.

WFO provides therapeutic recommendations in three categories: recommended, for consideration, and not recommended. Data were retrospectively analyzed to compare the postoperative treatment recommendations of MTB and AI. MTB recommendations were defined as concordant with WFO if they corresponded to the recommended or for consideration categories and as non-concordant if they corresponded to the not recommended or not available categories. Therapies recommended by the tumor board were classified as ‘not available’ if they were not known to the WFO at the time of analysis.

Data Analysis And Statistics

All statistical analyses were performed using SPSS (version 22.0; IBM Corporation, Armonk, NY, USA), with a significance level set at p < 0.05. Concordance was expressed as a percentage agreement. Continuous variables are expressed as mean ± standard deviation (range) and compared using a two-sample t-test. Categorical variables were expressed as frequencies and analyzed using the chi-square test. A logistic regression model was used to simultaneously control for the determinants of concordance, and odds ratios and 95% confidence intervals were calculated. Multivariate analysis was performed using Cox proportional hazards model.

The mean age of the 322 patients with gastric cancer was 64.43 years (SD, 12.42). Among them, 285 (88.5%) patients were under 80 years of age and 37 (11.5%) were over 80 years old. There were 227 men (70.5%) and 95 women (29.5%). The distribution of gastric cancer stages was 163 (50.6%) in stage I, 45 (14%) in stage II, 66 (20.5%) in stage III, and 48 (14.9%) in stage IV (Table 1). The Concordance rate between WFO and MTB was 72.98% at the recommended level (235/322) and 86.96% at the consideration level (280/322) (Fig. 1).

In univariate analysis, there were significant differences in age (p = 0.03), sex (p = 0.016), performance status (p = 0.012), tumor location (p = 0.014), and gastric cancer stage (p < 0.001) between the concordance and non-concordance groups. In contrast, there was no significant difference in HER2 status (p = 0.558), lympho-vascular invasion (p = 0.224), perineural invasion (p = 0.468), serum bilirubin level (p = 0.408), and serum creatinine level (p = 0.251) between the concordance and non-concordance group (Table 2).

Table 2

Comparison by concordance/ Univariate analysis of the concordance by variables.
Variables		Concordance (n = 280)	Non-concordance (n = 42)	P-value
Age	Total	63.71(12.22)	69.21(12.81)	0.007*
	< 80	252(90.0%)	33(78.6%)	0.030*
	≥ 80	28(10.0%)	9(21.4%)
Sex	Female	76(27.1%)	19(45.2%)	0.016*
	Male	204(72.9%)	23(54.8%)
Performance status	0	205(73.2%)	22(52.4%)	0.012*
	1	34(12.1%)	10(23.8%)
	2	23(8.2%)	3(7.1%)
	3	18(6.4%)	7(16.7%)
Tumor location	Proximal	71(25.4%)	9(21.4%)	0.014*
	Middle	62(22.1%)	18(42.9%)
	Distal	147(52.5%)	15(35.7%)
HER2 status	Negative	252(90.0%)	39(92.9%)	0.558
	Positive	28(10.0%)	3(7.1%)
Lympho-vascular invasion	No	188(69.9%)	12(57.1%)	0.224
	Yes	81(30.1%)	9(42.9%)
	Not available†	32
Perineural invasion	No	187(69.5%)	13(61.9%)	0.468
	Yes	82(30.5%)	8(38.1%)
	Not available†	32
Serum bilirubin	Total	0.58(0.50)	0.52(0.38)	0.490
	< 1.2	265(94.6%)	41(97.6%)	0.408
	≥ 1.2	15(5.4%)	1(2.4%)
Serum Creatinine	Total	0.73(0.18)	0.85(0.42)	0.090
	< 0.9	210(75.0%)	28(66.7%)	0.251
	≥ 0.9	70(25.0%)	14(33.3%)
Cancer stage	Ⅰ	158(56.4%)	5(11.9%)	0.000*
	Ⅱ	40(14.3%)	5(11.9%)
	Ⅲ	60(21.4%)	6(14.3%)
	Ⅳ	22(7.9%)	26(61.9%)
Abbreviation; HER2: Human epidermal growth receptor 2
* Statistically significant at p < 0.05.
Not available†: Data from patients with palliative chemotherapy.

More than half of the concordance group (56.4%) had stage I disease, and the concordance rate for stage I was the highest at 96.93%. The concordance rates for stages II and III were 88.89% and 90.91%, respectively, which were close to 90%; however, the concordance rate for stage IV was the lowest at 45.83% (Fig. 2).

The selected treatment options for each stage of gastric cancer were as follows (Fig. 3): All patients with stage I to stage III gastric cancer underwent curative gastrectomy. Among 163 gastric cancer patients with stage I disease, MTB selected surveillance for 160 patients and S-1 adjuvant chemotherapy for three patients. The selected treatment options for the 45 stage II gastric cancer patients were as follows: surveillance for four patients, S-1 adjuvant chemotherapy for 33 patients, and XELOX adjuvant chemotherapy for 8 patients. The selected treatment options for 66 patients with stage III gastric cancer were XELOX adjuvant chemotherapy for 37 patients and S-1 adjuvant chemotherapy for 29 patients. Of the 48 patients with stage IV gastric cancer, 32 patients were treated with palliative chemotherapy, of which 21 cases were treated with palliative S-1 + cisplatin chemotherapy, and 11 cases with palliative FOLFOX chemotherapy. The remaining 16 patients with stage IV gastric cancer underwent adjuvant chemotherapy after gastrectomy, and MTB selected adjuvant FOLFOX chemotherapy for 11 patients and adjuvant S-1 + cisplatin chemotherapy for 5 patients.

Table 3 indicates the results of the multivariate analysis of concordance as a function of age, sex, performance status, tumor location, and cancer stage. Several variables, such as age, performance status, and stage IV gastric cancer, had a significant effect on the concordance between MTB and AI (Table 3)

Table 3

Multivariate analysis of concordance using a binary logistic regression model.
Variables		OR	95% CI for OR	p-value
Age (ref: <80)	≥ 80	0.175	0.069, 0.441	0.000*
Sex (ref: Female)	Male	2.006	0.916, 4.391	0.082
Performance Status (ref: 0)	1	0.203	0.072, 0.574	0.003*
	2	0.191	0.057, 0.639	0.007*
	3	0.089	0.026, 0.301	0.000*
Tumor location (ref: distal)	Middle	1.058	0.377, 2.969	0.914
	Proximal	0.778	0.311, 1.943	0.590
Cancer stage (ref: Ⅰ)	Ⅱ	0.367	0.111, 1.220	0.367
	Ⅲ	0.673	0.218, 2.078	0.673
	Ⅳ	0.017	0.005, 0.055	0.017*
Abbreviation; OR: odds ratio; CI: confidence interval
* Statistically significant at p < 0.05.

This retrospective cross-validation study was conducted to investigate the factors that should be supplemented in AI to assist MTB in postoperative gastric cancer treatment recommendations. Only few studies have analyzed the concordance rate between AI and MTB in postoperative gastric cancer treatment recommendations. Choi et al. reported that stage IV gastric cancer was the only significant factor (6). In contrast, Tian et al. reported that HER-2 positivity was a significant factor, while stage IV was not (8). Surprisingly, in the present study, HER-2 positivity was not significant, but age > 80 years, performance status, and stage IV gastric cancer were all significant factors affecting the concordance rate between MTB and AI (Table 3).

Recommendations pertaining to patients aged > 80 years were less likely to be concordant than those pertaining to patients aged < 80 years (OR 0.175, 95% CI, 0.069–0.441; p = 0.000). Elderly patients have more comorbidities than younger patients, a tendency to refuse chemotherapy requiring hospitalization, and fluctuating general status (9, 10). These clinical features could have lowered the concordance rate in patients over 80 years of age.

The higher the performance score, the lower the possibility of concordance between MTB and AI in postoperative gastric cancer recommendation (OR 0.203, 95% CI 0.072–0.574, p-value = 0.003 for performance score 1, OR 0.191, 95% CI 0.057–0.639, p-value = 0.007 for performance score 2, OR 0.089, 95% CI 0.026–0.301, p-value = 0.000 for performance score 3). In other words, as the performance status score increases, it becomes easier to select a chemotherapy regimen that does not match the AI recommendation. In general, elderly patients have higher performance scores than younger patients (11).

In this study, there was no significant difference in concordance rates between stage II and stage III when compared to stage I (p = 0.367 for stage II, p = 0.673 for stage III). Interestingly, the concordance rate for stage IV was significantly lower than that in stage I as the preference for palliative chemotherapy regimen differed because of the difference in the local guidelines between MTB and WFO (OR 0.017, 95% CI 0.005–0.055, p = 0.017). S-1 plus cisplatin is a commonly used regimen in Korea and Japan, following the Japanese guidelines (12, 13). On the other hand, S-1 is an investigational agent in the NCCN guidelines of 2018, so it is not used in the Memorial Sloan Kettering Cancer Center (MSKCC) following the NCCN guidelines (14). Therefore, since S-1 + cisplatin was not included in the WFO chemotherapy regimen based on the MSKCC data, the concordance rate between MTB and AI was significantly lower in stage IV.

The discrepancies between MTB and AI for each gastric cancer stage and their reasons are summarized in the supplementary tables (Supplementary tables S1, S2, S3, and S4).

In stage I, there were five patients in the non-concordance group. AI recommended adjuvant chemotherapy, but MTB selected surveillance as the treatment option after gastrectomy. Unlike in the West, D2 lymph node dissection is commonly performed in East Asia; thus, observation is recommended in the Japanese guidelines for pathologic stage I after curative gastrectomy (13). MTB judged that since all of them underwent D2 lymph node dissection and were over 65 years of age, there was little benefit from adjuvant chemotherapy considering the complications of chemotherapy.

In stage II, five patients belonged to the non-concordance group. For four of them, AI recommended adjuvant chemotherapy, but MTB selected surveillance as a treatment option, considering their age, performance status score, and comorbidities. They were 68, 85, 69, and 72 years old, respectively, and their performance status scores were all grade 3. They also had comorbidities, such as dementia, chronic kidney disease, and chronic heart failure. For the remaining 1 gastric cancer patient, AI recommended FOLFOX as adjuvant chemotherapy, but MTB selected S-1 adjuvant chemotherapy because of the inconvenience of frequent hospitalization due to an old age of 82 years and a performance status score of grade 2.

In stage III, there were six patients in the non-concordance group. For four of them, AI recommended 5-FU or FOLFOX, but MTB selected S-1 adjuvant chemotherapy considering that they were all over 80 years of age, had a performance status of 2 or 3, and frequent hospitalization was inconvenient for them. For the other two patients, AI recommended S-1, capecitabine + radiation, or capecitabine + cisplatin, but MTB selected the XELOX regimen considering that they were young (49 and 64 years old, respectively), had a performance status of 1, and had no comorbidities.

In stage IV, 26 patients with the S-1 plus cisplatin regimen belonged to the non-concordance group.

AI is applied in various areas of medicine, such as robotics, medical diagnosis, and medical statistics. WFO, an AI system for clinical decision support, is expected to have many advantages, such as increased work efficiency and decreased workload of doctors, decision support for junior oncologists, and treatment selection based on the latest medical research, even in hospitals with few or no experts (15, 16). However, several factors lowered the concordance rate between the medical AI and experts in gastric cancer, resulting in the reduced validity of the medical AI.

First, AI lack a comprehensive understanding of individual patient. The WFO cannot understand the comprehensive status of patients, such as patient compliance and rapport with doctors, comorbidities that may affect chemotherapy, and interpretation of whether biochemical study results are temporary or persistent (5).

However, it is expected that these shortcomings of AI will be compensated for as the technology advances. A wearable device or sensor can check the patient's condition 24 h a day and evaluate the patient's activity. In this way, continuous rather than fragmentary information can be obtained, and accurate individual performance status can be obtained through individual activity history rather than performance scores classified into scores such as 0, 1, 2, and 3 (17, 18). The development and usage of health applications that can be used in portable computer devices such as smartphones are also expanding. If the medical information recorded in a personal device application can be easily linked to the medical information database of a country or hospital, the patient's medical history and comorbidities can also be easily identified and applied in clinical practice (19, 20). In addition, AI assistance for emotional support, such as a robot companion for the elderly with limited cognitive function or activity, is expected (21).

Second, the local guidelines for gastric cancer differ according to race, country, and region. WFO, the medical AI used in this study, is based on the data of MSKCC in the US, but gastric cancer treatment in Korea follows the Korean guidelines, which is closer to the Japanese guidelines than the NCCN guidelines (12). Local guideline differences are caused by differences in preferred surgical methods, the effectiveness of specific chemotherapy regimens or radiation therapy, and the approval status of chemotherapy drugs by country. In the future, besides resolving these local guideline differences, AI may assist in determining the best treatment plan tailored to individual patient with cancer.

Third, there are several economic factors. Owing to the high cost of cancer medicines, an individual patient’s financial circumstances and coverage of public or private insurance affect the selection of the chemotherapy regimen. Therefore, lowering the price of chemotherapy drugs and expanding insurance coverage will enhance the affordability and accessibility of cancer medicines (22). AI increases the efficiency of clinical trials and research, thereby significantly lowering the cost of drug development, which in turn lowers the price of cancer medicine (23). The increase in the efficiency of chemotherapy drugs is also expected to have a positive impact on social discussions and government approval regarding the insurance coverage of anticancer drugs.

This study had several limitations. First, this study was a retrospective and single-center analysis; therefore, it may have been biased. Second, the results of this study were analyzed based on the treatment consensus for WFO and MTB from 2015 to 2018. If we compare the concordance rates between the last version of WFO and MTB based on the latest guidelines, the results may differ. However, because S-1 is still considered an investigational agent in the NCCN guidelines, it is most likely that there will be no change in the concordance rate for stage IV. In addition, the NCCN guidelines for gastric cancer, version 2022, also recommend adjuvant chemotherapy if there is LN metastasis after gastrectomy. Thus, the stage I discordant groups will still be discordant. An additional second-blind test for the remaining discordant group could not be performed because of changes in the MTB members. Third, this study analyzed only the concordance rate as a method for evaluating the validity of AI. It has not been analyzed to what extent AI influences doctors’ decisions, whether better outcomes such as increased overall survival or disease-free survival can be achieved with AI-recommended treatments, and how much time and cost can be saved by using AI. These factors should be analyzed not only for the validation of AI but also in terms of the usefulness and economic feasibility of AI. However, to analyze these factors, a large-scale prospective study is required, and discussions on the ethics and legal responsibility of AI decisions should be conducted before such a study.

In this study, the factors affecting the dis-concordance between AI and MTB were age, performance status, and stage Ⅳ gastric cancer. The effect of gastric cancer stage IV occurred because of the difference in the local guidelines between AI and MTB. And the effects of age and performance status were caused by the absence of AI’s the ability to comprehensively understand individual patients.

By taking into account the local cancer guidelines and developing the ability to understand the comprehensive status of cancer patients, a medical AI may be able to not only effectively assist MTB in the management of gastric cancer patients in actual clinical practice, but also play a role as a member of MTB.

AI, Artificial Intelligence; MTB: Multidisciplinary tumor board; WFO: Watson for Oncology; DCMC: Daegu Catholic University Medical Center; HER2: Human epidermal growth receptor 2; OR: Odds ratio; CI: Confidence interval; MSKCC, Memorial Sloan Kettering Cancer Center.

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of the Daegu Catholic University Medical Center in Daegu, Republic of Korea (IRB No. DCMC-CR-21-017).

Consent for publication

Not applicable

Availability of data and materials

The anonymized data used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

The authors declare that they have no funding.

Author’ contributions

YEP and HDC designed the study. YEP collected the data. YEP and HDC analysed and interpreted the patient data. The HDC commented on the drafts of the paper. YEP was a major contributor to the writing of the manuscript. All authors have read and approved the final manuscript.

Acknowledgements

Not applicable.

Berardi R, Morgese F, Rinaldi S, Torniai M, Mentrasti G, Scortichini L, et al. Benefits and Limitations of a Multidisciplinary Approach in Cancer Patient Management. Cancer Manag Res. 2020;12:9363–74.
Curioni-Fontecedro A. A new era of oncology through artificial intelligence. ESMO open. 2017;2(2).
Mirbabaie M, Stieglitz S, Frick NR. Artificial intelligence in disease diagnostics: A critical review and classification on the current state of research guiding future direction. Health and Technology. 2021;11(4):693–731.
Tang X. The role of artificial intelligence in medical imaging research. BJR| Open. 2019;2(1):20190031.
Jie Z, Zhiying Z, Li L. A meta-analysis of Watson for Oncology in clinical application. Scientific reports. 2021;11(1):1–13.
Choi YI, Chung J-w, Kim KO, Kwon KA, Kim YJ, Park DK, et al. Concordance rate between clinicians and Watson for oncology among patients with advanced gastric cancer: early, real-world experience in Korea. Canadian Journal of Gastroenterology and Hepatology. 2019;2019.
Suwanvecho S, Suwanrusme H, Sangtian M, Norden AD, Urman A, Hicks A, et al. Concordance assessment of a cognitive computing system in Thailand. American Society of Clinical Oncology; 2017.
Tian Y, Liu X, Wang Z, Cao S, Liu Z, Ji Q, et al. Concordance between Watson for oncology and a multidisciplinary clinical decision-making team for gastric cancer and the prognostic implications: retrospective study. Journal of Medical Internet Research. 2020;22(2):e14122.
Lichtman SM, editor Chemotherapy in the elderly. Seminars in oncology; 2004: Elsevier.
Radecka B, Czech J, Siedlaczek A, Maczkiewicz M, Jagiełło-Gruszfeld A, Duchnowska R. Chemotherapy compliance in elderly patients with solid tumors: a real-world clinical practice data. Oncology in Clinical Practice. 2022.
Berthelot G, Johnson S, Noirez P, Antero J, Marck A, Desgorces F-D, et al. The age-performance relationship in the general population and strategies to delay age related decline in performance. Archives of public health. 2019;77(1):1–9.
Guideline Committee of the Korean Gastric Cancer Association DWG, Panel R. Korean practice guideline for gastric cancer 2018: an evidence-based, multi-disciplinary approach. Journal of gastric cancer. 2019;19(1):1–48.
jp JGCAjkk-ma. Japanese gastric cancer treatment guidelines 2014 (ver. 4). Gastric cancer. 2017;20(1):1–19.
Qiu H, Zhou Z. Updates and interpretation on NCCN clinical practice guidelines for gastric cancer 2017 version 5. Zhonghua wei chang wai ke za zhi = Chinese journal of gastrointestinal surgery. 2018;21(2):160–4.
Murphy EV. Clinical decision support: effectiveness in improving quality processes and clinical outcomes and factors that may influence success. The Yale journal of biology and medicine. 2014;87(2):187.
Printz C. Artificial intelligence platform for oncology could assist in treatment decisions. Cancer. 2017;123(6):905-.
Lu L, Zhang J, Xie Y, Gao F, Xu S, Wu X, et al. Wearable health devices in health care: narrative systematic review. JMIR mHealth and uHealth. 2020;8(11):e18907.
Wu M, Luo J. Wearable technology applications in healthcare: a literature review. Online J Nurs Inform. 2019;23(3).
Hong L, Luo M, Wang R, Lu P, Lu W, Lu L. Big Data in Health Care: Applications and Challenges. Data and Information Management, 2 (3), 175–197. 2018.
Lee J, Kim H-S, Kim J. Out-of-hospital data: patient generated health data. The Journal of Korean Diabetes. 2020;21(3):149–55.
Abdollahi H, Mollahosseini A, Lane JT, Mahoor MH, editors. A pilot study on using an intelligent life-like robot as a companion for elderly individuals with dementia and depression. 2017 IEEE-RAS 17th International Conference on Humanoid Robotics (Humanoids); 2017: IEEE.
Organization WH. Technical report: pricing of cancer medicines and its impacts: a comprehensive technical report for the World Health Assembly Resolution 70.12: operative paragraph 2.9 on pricing approaches and their impacts on availability and affordability of medicines for the prevention and treatment of cancer. 2018.
Ocran Mattila P, Ahmad R, Hasan SS, Babar Z-U-D. Availability, affordability, access, and pricing of anti-cancer medicines in low-and middle-income countries: a systematic review of literature. Frontiers in Public Health. 2021:462.

No competing interests reported.

SupplementaryTables.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Experience and future perspective of artificial intelligence (AI) in postoperative gastric cancer treatment recommendation: a retrospective, observational, cross-validation study with multidisciplinary team board

Status:

Version 1

Abstract

Figures

Introduction

Methods

Patients and study design

Ai And Mtb Concordance Determination

Data Analysis And Statistics

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1