What are your suggestions

[Wirichada Pan-ngum]

What are your suggestions on coping with those challenges? (10 marks)

—————————- Deadline 23 January 2022 Pls reply before ————————————

[Zarni Lynn Kyaw]

From the paper the following challenges are identified –
1) The vast amount of data that is generated from various sources (such as electronic health records, social media, wearable devices, etc.) can be difficult to handle and analyze.

2) The quality and reliability of data may vary depending on the source

3) The data can be complex and multi-dimensional, making it difficult to extract relevant information

4) There may be a lack of standardization in the data, making it difficult to compare or combine information from different sources

5) There may be privacy and security concerns related to the collection and use of big data

6) The integration of big data into clinical care can present challenges for medical professionals in terms of data interpretation and integration into clinical decision-making

7) Limited resources and knowledge may make it difficult to fully take advantage of the potential of big data in cardiovascular research and clinical care.

Let me use Myanmar as a case to share my views on how to cope with those challenges –
a in Myanmar:

1) Data management and governance: Develop a robust data management and governance infrastructure that includes data storage, security, and privacy protocols to ensure the safe and secure handling of data. This will also help to ensure the quality and reliability of the data.

2) Data analysis and visualization: Utilize advanced data analysis and visualization tools to make sense of the vast amount of data. This can include machine learning, natural language processing, and data visualization techniques to help extract relevant information and make it more understandable for decision-makers.

3) Data standards and interoperability: Establish data standards and interoperability protocols to facilitate the sharing and integration of data from different sources. This can help to overcome the lack of standardization and make it easier to compare and combine information from different sources.

4) Training and capacity building: Invest in training and capacity building for health professionals and researchers to help them navigate the complexities of big data and use it effectively.

5) Collaboration and partnerships: Develop partnerships and collaborations with other stakeholders, such as universities, research institutes, and private companies, to share expertise and resources and help to address the challenges of big data in Myanmar.

6) Addressing Digital divide : with the internet connectivity being limited in certain area and digital literacy being low in certain group of society, it is important to address digital divide and empower those who are at disadvantage. This could include providing digital literacy training, increasing internet connectivity in under-served areas, and establishing digital health clinics.

7) To address the challenge of privacy and security concerns, one suggestion would be to implement robust security and privacy protocols to protect the data from unauthorized access or breaches. This can include measures such as data encryption, secure data storage and access controls, and regular data security audits.

8) To address the challenge of data interpretation and integration into clinical decision-making, one suggestion would be to provide training and education for medical professionals on the use of big data in clinical care, including data interpretation and integration into clinical decision-making.

9) To address the challenge of limited resources and knowledge, one suggestion would be to collaborate with other organizations and institutions, such as universities and research centers, to leverage their expertise, resources, and infrastructure. It may also be beneficial to explore external funding options and government funding opportunities to support big data research and implementation in Myanmar.

[Siriphak Pongthai]

This is good to know how you cope those challenges in your country. Thank you for sharing those useful information 🙂

[ABDILLAH FARKHAN]

Appreciate that you have shared comprehensive approach in responding to the challenges in the big data era. While I pay too much attention on human resource capacity, other focuses such as collaboration, partnerships and governance are also important. It cannot be denied that the development of big data and AI may be faster than what humans can do.

[PREUT ASSAWAWORRARIT]

Although there are great benefits and opportunities from big data, these make it great challenges to informaticians to retrieve specific data, process them, and transform them into valuable information or clinical applications. The following are such challenges and my suggestions on how to cope with them.

1. Multiple definition. There are some diseases or terms that have synonyms or can be written in other words. We have to think of all possible synonyms that have been used. Moreover, if the terms are unstructured data, we will face with the term which was misspelled and could not be tracked by searching the word. In this scenario, redundant matching of terms can be used to detect some misspelled words.

2. Unstructured data. In many hospitals, many data is still in unstructured form, for example, scan documents, images, history, physical examination, progress note, nurse note, etc. Transformation from unstructured to structured one is essential. However, we have to set up the data that need to be transformed. In addition, some data are found in scan documents written by difficult-to-read hand-writers.

3. Missing data. Every database has missing data which most of them often do not miss at random, leading to selective bias. If the missing data do not exceed 10 percent of all data, there are statistical methods to solve the problem. The methods include imputation techniques, mixed effects regression model, generalized estimating equations, and inference technique. However, increasing the proportion of missing values can lead to compromised results.

4. Data inconsistency. Inconsistencies in the data may occur after we duplicate the data. We have to check the consistency of the data by checking them.

5. Clinical applicability. There is evidence that most of the results analyzed from the big data are not true after well-designed randomized controlled trials are conducted. The results derived from a research on big data tell us about a trend that needs to be confirmed with a standard randomized controlled trial.

6. Legal and ethical issues. The privacy and confidentiality of patients are our main concerns. All identifiable data must be encrypted to de-identify the patients. Additionally, a password may be needed to access the database.

[Boonyarat Kanjanapongporn]

Thank you for sharing idea to deal with big data. I do agree with you especially with the using of RCT to confirm the result from big data. This would enhance the accuracy of data application by control the factors which are hard to be managed previous from big data such as bias or missing value.

[Kansiri Apinantanakul]

Thank you for sharing ka.
I agree with you that the multiple definition and the misspelled word, as well as the data discrepancy is the concern to be addressed.
I have the experience of matching the different trade name of the drug within generic name.
We could also solve this issue by grouping and re-coding but it could take a long time.

[Zarni Lynn Kyaw]

Thanks for your replies. I’m now exploring possibilities to use tools like Unreal engine 5.1 where I can input the data and explore the data in 3D. I learned that it is possible to do it at JITMM 2023. I’m saving up for a 3D goggle and maybe manipulate the data in VR 🙂

[ABDILLAH FARKHAN]

Big data in cardiovascular disease (CVD) is leverage to improve treatment, research, and development of CVD. However, management of big data contains significant issues where capacity of human resources (health informaticians or data scientists) is the key to addressing any significant complexities. Here are some practices of coping mechanisms when dealing with challenges of big health data:

a. Handling any missing data: although this issue can be overcome by using statistical approaches (imputation, regression, estimation, and inference), understanding how to select those methods must be done carefully to mitigate the high number of missing values.

b. Avoiding selection bias: it is about how researchers defend their principles when are intervened by data and science. Avoiding selection bias is a difficult challenge, but it is true that large volumes of big data do not always imply a representative sample with any valid inference. The electronic health record of CVD patients is likely to store exposure history and risk factor information that is scientifically related to CVD, but these cannot be generalized for decision-making. In clinical research, randomized controlled trials have the greatest influence.

c. Mainstreaming ethical clearance and data security (protecting the subject’s personal information from malicious actions): implement internet and medical data encryption, multilayer authentication, firewalls, and administrative safeguards under the principle of data confidentiality-integrity-availability.

d. Integrating big data management into healthcare workforce education: apart from becoming a challenge, big data is actually an opportunity for higher education institutions to prepare digital-literate health workers. Education and training curricula must contain big data applications so that every health worker can be ready to actualize clinical service in line with the nowadays demands.

e. Interpreting the correct translation: huge volume, variety, and velocity of big CVD data enables multiple assumptions. Once the users play with big data, they should be able to execute visualization and text mining in order to provide the appropriate translation.

[Boonyarat Kanjanapongporn]

Big data in healthcare would possibly improve quality of care for large scale, varieties of data being included and recorded timely. Even though many benefits would be received, complexity of using big data is an obstacle in achieving effective outcomes. There are many obstacles from data sources, data processing and data usage. Some problems would probably be predicted and some might not. Below are my ideas to deal with the barriers of using big data.

First, Dealing with technology and data system adjustment.
There are many problems which could probably predicted or managed by doing Risk assessment and Operating plan to adjust the system. From the article, different disease definitions, unstructured data and data linkage are mentioned and these issues might be resolved by using system with interoperability standard. Moreover, moving the silos system to cloud network might increase the ability of data sharing between investigators. Data inconsistency and Data integrity could also be protected by applying Information security methods such as Malware protection, Access control and Encryption. There might be many other technical problems but data system management could assist with smoothing the process.

Second, Application considerations.
Insight generated by big data needs to be considered before applying. High percentage of Missing value and Bias are the examples of data which needed to have cautious usage. Large and varieties of data collection might not suit some specific and desirable clinical situations for application. Therefore, Interpretation results would need tracing back to credible sources and methods of information generation.

Third, Human training.
Without human training, the problem in any aspects of technology and data, might not be easily solved. Time and money might have to be invested to build strength in Human Resources for the proper data system management, interpretation and problem solving. Moreover, Legal and ethical issues might have to be reminded among staff, and used with sensitive and confidential data. For example, HIPAA could be one of the regulations where staff involved should be acknowledged and comply.

Overall, there are varied problems and methods to handle it. It’s probably too idealistic to be able to manage all the obstacles without consideration on limited resources. Therefore, Prioritizing the problem might assist the staff to arrange order of important issue under the real situation with limited budget and Human Resources.

[Kansiri Apinantanakul]

Thank you for sharing.
I agree with you that the well-trained staff in wide range of aspects including the laws and regulations, data literacy, technology literacy is the core foundation of in big data era.

[Tanyawat Saisongcroh]

Totally agree with you mentioning prioritize issues. In term of training researcher or clinician in data science management, particular in big data and machine learning, it might be a big challenge as it needs quite time and experience to be an expert. Investment in hiring data analyst professionals might probably be an effective option.

[Siriphak Pongthai]

First of all, this paper is very good in mentioning how useful and important of big health data. Particularly when they mentioned that stent thrombosis in patient with CAD undergoing PCI was found through large-scale studies.

There are many challenges for research in using big health data indicated in this paper:
– Disease definition/ Unstructured data: disease definition may be varied among countries. By using clinical coding standards (such as ICD-10) in classifying diseases and health problem could help in heterogeneous of disease classification. Yet, the data from each EHR and sources can be different and could cause big obstacles in collecting and analyzing of data. Therefore, interoperability standards should be set among systems, hospitals, and stakeholders. This will ease data transfer and management.

– Legal and ethical issues: especially when many countries have started to be enacted PDPA just like Thailand, or HIPPA rules which was legislated in the States. This could be difficult for future research in using patient’s data in the past. I would suggest that hospital should have agreement at first by asking for consent in sharing medical data for example, laboratory results or imaging results etc. In addition, we must make sure that patient identifiable information must be blinded prior to sharing. Nonetheless, the ethical committee should be in part in considering what research can or cannot do.

– Data security/ Data integrity: researcher team must have specialized IT staff in managing information and system security. The data system should have policy of access control and audit trail to make sure that data are complete, consistent, and accurate to original. Furthermore, backing up data method is crucial to prevent data lost.

– Data quality and missing data/ Data inconsistency/ Training: researchers must be well trained to make sure that data are completely and correctly collected. If we implement interoperability standards at first, we will have a set of information that must be collected from subjects. Thus, this helps in prevent missing of data. In addition, our lives would be easier by having IT helps in checking consistency of data. Most importantly, to solve missing data and ensure quality of data, researchers must also have competence and knowledge in data analytics with multiple testing, in order to know which methods could solve those complexities.

There are many more solutions and suggestions for each challenge mentioned above. However, some of them might or might not be applicable depend on real world setting.

[Kansiri Apinantanakul]

Thank you for sharing, your comment on the PDPA is interesting.
I have a few things to add regarding the broad consent.
I think the consent for PDPA is important, but I also heard about the broad consent that we should take in consideration as well.

[Kansiri Apinantanakul]

Hi All,

Below is my suggestion on each challenge.
Missing data:
I agree with the paper that most of the data is not missing at random. It caused by the standard practice or patient refusal. If the proportion of the missing is quite low, I would personally start with the completed case analysis. If the proportion is relatively high, I would try the data analytic technique, for example, imputation. However, I would always keep in concern that the imputed data is not real one. The imputation could lead to the analytical bias. If there is the alternative datapoint that could be used and that datapoint is more completed, I would try that also.

Selection Bias:
As this paper mentioned about the selection bias due to the nature of data collection, I do agree with this statement. The selection bias could not be avoided when we obtained the data since we did not create it or collect it ourselves with the standard method. In my practice, I would perform preliminary test to get the overview of the data first. The basic statistical parameter should be calculated as well as the study of data distribution. This could give the analyst the big picture of the obtained data and sometimes we may see the weird pattern due the selection bias as well.

Data analysis:
The limitation of knowledge on big data analytic and the algorithm developed to handle big data has been discussed for a while. I suggestion that the researcher should be trained on the handle of big data along with the statistics. The refreshing training and update on the newly released algorithm should be provided on the regular basis.

Applicability of the results:
For this issue, I suggest that the analyst should provide the result and ensure the data processing transparency as much as possible. The complex algorithm is not generally acceptable in healthcare field since most of healthcare staff has little to no data literacy. The complex algorithm is the “black box” for them. In term of the reader, all of us should aware of the big data trend and the important of data literacy that we should seek for.

Privacy and Ethical issue:
For this issue, I suggest that the data owner should prioritize the data privacy and the data security on the top of all things. Data is the asset. The data owner should be aware of it and invest on the data security measures. Apart from that, the data owner should try their best to comply with the local law and regulation.

[Kawin Wongthamarin]

From my point of view, I think the way to deal with each challenging topic is as follows.

Missing data
Finding missing data is an unavoidable aspect of working with data, and it is important to have a plan in place to systematically handle it. While this may be manageable for smaller scale surveys, it can become a significant challenge when dealing with large datasets or data from electronic health record systems. To effectively address this issue, it is crucial for informaticians to be educated on the various methods for coping with lost data. In addition, Data validation and quality checking at the data entry stage will help reduce the amount of missing data.

Selection bias
Despite the potential for selection bias, observational research can be a valuable starting point for exploring new possibilities and understanding complex phenomena. It can provide important insights and serve as a foundation for more in-depth studies using methods such as randomized controlled trials or molecular experiments. Additionally, observational research is often more cost-effective and less time-consuming than other types of studies, making it a useful tool for researchers to utilize in the early stages of their research. So I think the way to deal with selection bias is important to be aware of this bias and understand it in order to accurately interpret and make decisions based on the data.

Data analysis and training
With the constant emergence of new data analysis methods, it is important to be familiar with multiple approaches. No single method is suitable for all types of data, and each has its own strengths and weaknesses. Having a diverse knowledge of analytical methods increases the chances of successful data analysis.

Privacy and ethical issues
In order to cope with this problem, I think that it is crucial for the government to establish clear and comprehensible laws and to provide researchers with the necessary training in information security. This will ensure that research can continue in a safe and secure manner.

[Tanyawat Saisongcroh]

There are several challenges mentioned in the article. In my opinion, there are 4 main points that should consider to cover those issues, whether organization level or governance level

(1) Recruit skilful professionals
– Big data analyst professionals
– Cybersecurity professionals
– IT professionals

(2)Increasing training program for everyone in work environment
– Those professionals could held workshop and seminar for everybody at least to proper handling data regularly, knowledge in storage, importance of data management
– Particular training in big data handling for clinicians or researchers to gain more skills
– Practice in cybersecurity; data encryption, identification and access authorization control, real-time monitoring, endpoint security

(3)Invest in digital technology, for example
– Software automation tools or knowledge analytics solution powered by AI/Machine learning to handle big data to help in data analysis and interpretation
– System security maintenance such as Big data security tools (IBM Guardium), endpoint security implementation

(4)Conduct proper research methodology with well-designed statistical adjustment to prevent or reduce bias
– Review the literature carefully in order to specify more proper scientific subjects, related dataset variables and sources
– Prioritize in using randomization selection
– Consult the expert for addressing the confounding with several advance statistical techniques
– Use software automated tools, which contain pre-built APIs for a broad data spectrum

[Hazem Abouelfetouh]

In my opinion, We can implement several strategies to face these challenges including:

Disease definition: The quality and comparability of big health data can be enhanced by developing standardized and consistent disease definitions.

Data availability: Big health data can be made more accessible by forming alliances with a variety of data sources by developing partnerships to increase data availability with various data sources, such as hospitals, clinics, and research institutions. Also by involving cloud-based solutions and platforms to store and share large amounts of data, making it more accessible for remote teams and researchers.

Data quality and missing data: Data quality can be improved and missing data can be addressed using methods like data validation to check the data for any errors or inconsistencies, imputation by using statistical techniques to fill in missing data based on the observed data, and data cleaning to identify and remove errors, outliers, and inconsistencies in the data.

Data inconsistency and Data integrity: Data inconsistency can be addressed by developing methods for integrating and standardizing data from multiple sources. The integrity of large amounts of health data can be safeguarded by putting in place stringent data security and privacy measures.

Training: Developing specialized training programs for individuals working with big health data can help to ensure that they have the necessary skills and knowledge to work with this type of data.

Legal and ethical issues: Developing policies and guidelines to address legal and ethical issues related to big health data can help to mitigate risks and ensure compliance. And improve healthcare organizations’ transparency about how they collect, use, and share health data.

Data security: Implementing robust security measures such as encryption, firewalls, and network segmentation can help to protect big health data from unauthorized access and breaches. Data can be encrypted to protect it from unauthorized access or breaches.

Fairness and bias: Implementing fairness-aware machine learning methods and monitoring the data and models for any potential bias can help to mitigate any unfairness in the predictions.

Overall, it’s important to combine technical solutions such as encryption, access controls, network security, and data backup with policy, and governance, and collaborate with domain experts and stakeholders to ensure that the data and models are fair and align with ethical and social values.

[Tanatorn Tilkanont]

The challenges are in every step of Big Health Data management, starting from the source of data collection to its application. In this articles, Silverio A, et al. had mentioned the challenges of big health data and how to deal with those challenges, including;

Missing Data: The missing data in big health data is uncontrollable. Thus, this article suggested several methods for handling a large number of missing data, such as Imputation technique, Mixed effects regression model, Generalized estimating equation, and Inference. Several methods suggested to handle miss data that is less than 10% would remain the same distribution and prevent any outlier data.

Selection Bias: A large scale of health data from EHR would have many confounding factors as the data are from different sources, interventions, inconsistent, and independent. Big data is observational studies that reflect actual cases in the real world. I believe that the information from Big data analysis would be an important element that supports the randomized controlled trial design to confirm the hypothesis.

Data Analysis and Training: I personally believe that more clinicians and researchers are interested in training on big data analysis using appropriate statistical and methodological tools. However, only few researchers are able to interpret with more complex data. It would be great if clinical researchers themselves are trained with informatics, coding, data analysis, etc. The availability of well-trained researchers are required as advisory.

Data Privacy and Ethical Issue: Health data in Thailand are mostly centralized. The data confidentiality and privacy are primary concerns. The Personal Data Protection Act (PDPA) is implemented in Thailand to handle ethical issues. Patients provide their broad consent to allow using their health data on further research. However, we also need to ensure that our system security is good enough to protect data from cyberattackers.

[SIPPAPAS WANGSRI]

What are your suggestions on coping with those challenges? (10 marks)

From the research paper, there were 14 challenges identified in using big data, its application and implementation in the field of health care. There were some important points that this publication addressed, which were;

1. Missing Data – from my experiences in dealing with Thai EMR data (43 Files), missing data is inevitable. The reasons behind them were exactly as this research stated, because of the poor data collection system, omission by clinicians and health care providers including but not limited to nurses, residents and staffs. In this problem, I suggest to redesign data collection method and encourage users to aware of how entering a valid data will be beneficial to them (in research purposes, for instance). After that, there should be data validation process to help verify data quality and data integrity.
2. Selection bias — Because the data collected for big data analysis is primarily based on different hospitals with their own EMR database. To deal with the data biases, pooling data with different hospitals in a different region, or even a national scale would help reduce this problem.
3. Data analysis and training — this issue can be overcome by training a specialised profession in data science with the combination of knowledge in medical field and programming field.
4. Interpretation and translational applicability of results — Data gained from EMR, however, is retrospective data and will ultimately require another research methodology to validate the application for further use in clinical practice. Also, in data science with artificial intelligent models are somewhat known as the “black box” in which the data output can not be proven by the traditional method. For physicians in clinical practice, they often rely on the evidence-based and this new approach of big data analytics using machine learning model might not be easily provide them sufficient obvious evidence.
5. Privacy and Ethical issue — This is a serious concern, and in my opinion, the most problematic issue which causes a significant barrier in big data analytics in health care. Because big data possesses large volume and velocity characteristics, they require the cutting-edge computational resources and hence, we can achieve them with cloud computing. For legal reasons, including Thailand, the government prohibits storing government data (health care data, included) on the cloud due to fear of data being compromised. The only way we can cope with this issue is that we have to ensure that the data being uploaded to the cloud is strictly and follow the HIPAA and security regulations.

[Zarni Lynn Kyaw]

Hi All,

I’ve been joining PMAC and I would like to share an opportunity to join and learn about Planetary Health. The presenter convince me that we have to go beyond health to tackle the upcoming climate change challenge.

https://BIT.LY/SEACHRC

We can get a certificate after joining all the Zoom classes.

[Wirichada Pan-ngum]

Thanks for sharing the link. I was there on one of the pre-meeting workshop on modelling as well. It is such a well organised event!

[Zarni Lynn Kyaw]

It is a very well organized event but unfortunately my abstract was rejected because this year’s theme is mostly focused on climate change and many climate related projects in Myanmar was on paused after the military coup since 2021 and we didn’t have anything related to present this year.

[Wirichada Pan-ngum]

Great discussions among yourselves. I learn as much from the discussions here after reading the article. Thanks guys for your participation.

[Author]

Posts