Forum Replies Created
-
Posts
-
-
-
1. Why was the author interested in investigating the suicide problem in Thailand during the time?
– Within 2005 to 2014, the suicide rates in Thailand had been shockingly numbered at around 3,600 to 4,000 people every year. Also, the suicide rates had generally been rising since 2011, at more than 6 suicides per 100,000 inhabitants.
2. Pick one potential risk factor mentioned in the paper and explain how the variable can contribute to the suicide rate?
– FEMH (Female as the head of the household), or when look conversely, male as the head of the household is risk factor.
– After AGE60 (age > 60) and DIV (divorce), this can easily be the third biggest risk factor.
– When male counterpart is the head, the female are more prone to abuse by male; but not as much vice versa. This contributes to overall happiness.3. How statistical modeling can contribute to investigate the epidemiology and spatial aspects of Thai suicide problem?
– On the epidemiology itself, statistical modeling can identify risk factors, for example, with multiple linear regression.
– On the spatial aspect, this study in confined to a target population, in a specific time period. However, it didn’t tell much about clustering or hotspots. -
1. What are possible reasons locations in epidemiological research have not been incorporated as much as other components in epidemiological research? How can spatial epidemiology be considered as an interdisciplinary science?
– Location can be under-recognized for associations, without correct tools. And, even with the tools, analyzing spatial associations still contain many pitfalls. That’s why it is both difficult and had to be studied separately.
– Spatial epidemiology has to be studied, to discover more tools, and consider more of nitty-gritty of spatial research. Technological advances may also help, but had to be studied further.
2. Explain why it is widely recognized that the place where an individual lives or works should be considered as a potential disease determinant and give some examples?
– The place where an individual lives involves being nearer to the risk factors, which may also be an area or point estimate, where, the nearer, the more risky.
– For example, being closer to a factory causing risks in contacting chemical wastes. Or, being in a close quarter, where people migrate often, is a risk factor to human-human communicable diseases. However, one must also consider how risk factors travel. Airborne? Short-lived droplets? Eating and drinking?
-
Yes, and safeguarding is quite difficult too; unless you prevent authorized usage in the first place. Also, with data mining / AI techniques, more identifiable information are more likely to be found. Think reverse engineering…
– Year of birth: 1990
– Specialty: Primary care doctor
– Workplace: a district hospital in Nakorn Pathom Province, ThailandAnd, the problem exacerbates if sample size is smaller.
-
-
-
Perceived Ease-of-Use (EOS)
– Approachability of digital devices
– Computer literacy
– Intuitiveness of the interface
– Fast responsiveness / interactivity
– Responsive design (RD) to viewport / unpredictability of user’s device
– Buglessness / tolerance to unpredictability of user’s input
– Getting help – can get the right Q&A quickly, if possible; otherwise, prompt Q&A session with humans (or machines that is adaptive enough)Perceived Usefulness (U)
– Escaping from past – realizing that what used to work can be better (positive); or it was so bad (negative)
– Embracing the possibility – it would be better if, this way
– Inevitability of the future and competition
– Environment / peer pressure (herd mentality)
– Pressure from the authorities / bosses
– Lack of workarounds -
Efficacy refers to whether the action provides clinical results under ideal condition (well-controlled condition).
Effectiveness refers to whether it can achieve results under usual circumstances (average clinical conditions).
This article has a infographic that explained the difference well.
Efficiency, as I have found, seems least confusing of the three terms, and refers to value of the intervention itself, relative its cost on society; that is, Return On Investment / Cost of Ownership.
For efficiency, I am looking at PubMed search, in particular, this article.
-
#21 Page 344
If two confidence intervals overlap, the difference between two estimates or studies is not significant.
It still requires a statistics to compute difference between estimates, and still may produce a difference.
I don’t know if I understand correctly, but vice versa cannot be concluded as well.
It can, however, be noted that if the two 95 % confidence intervals fail to overlap, then when using the same assumptions used to compute the confidence interval, we will find < 0.05 for the difference; and if one of the95 % intervals contains the point estimate from the other group or study, we will find > 0.05 for the difference.
-
Are Patients With Cancer Less Willing to Share Their Health Information? Privacy, Sensitivity, and Social Purpose
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4575401/
– Main objectives of the study
Growing use of electronic health information increases opportunities to build population cancer databases for research and care delivery. Understanding patient views on reuse of health information is essential to shape privacy policies and build trust in these initiatives.
– Main exposure variable of interest
We randomly assigned nationally representative participants (N = 3,336) with and without prior cancer to six of 18 scenarios describing different uses of electronic health information. The scenarios varied the user, use, and sensitivity of the information.
– Main outcome variable of interest
Participants rated each scenario on a scale of 1 to 10 assessing their willingness to share their electronic health information.
– Limitations of the study
First, the study was not originally powered to detect differences between cancer versus noncancer participants. This limits our ability to test for more complex interactions. Second, we do not know when the respondents were diagnosed with cancer or the type of cancer. Patients with a current or recent diagnosis of cancer or patients with specific types of cancer with different levels of heritability might have different preferences regarding reuse of their health information. For example, cancers that are familial, more consequential, or occur at a younger age might yield different views about information sharing. Third, we tested only two different levels of sensitivity in our conjoint experiment. Had we included a broader range of possibilities, we might have found that sensitivity was more or less important. Nevertheless, the finding that cancer participants were more favorable toward reuse of their health information when genetic information is included is a novel finding. Fourth, we presented participants with hypothetical scenarios rather than observing real-world decisions. Therefore, we were unable to measure how actual changes in behavior would be correlated with effect sizes in our experiment. However, responses to hypothetical scenarios have been shown to be highly predictive of behavior. In addition, we used a controlled experimental design that is more likely to reveal individual preferences than static survey questions.
-
In the literature review related to your research topic, please select a publication that used cross-sectional study design. Then describe:
1. The title of the paper
Leveraging EHRs for Patient Engagement: Perspectives on Tailored Program Outreach Am J Manag Care. 2017 Jul 1; 23(7): e223–e230.
2. Main objectives of the study
Electronic health records (EHRs) present healthcare delivery systems with scalable, cost-effective opportunities to promote lifestyle programs among patients at high risk for type 2 diabetes, yet little consensus exists on strategies to enhance patient engagement. We explored patient perspectives on program outreach messages containing content tailored to EHR-derived diabetes risk factors—a theory-driven strategy to increase the persuasiveness of health communications.
3. Sampling method used in the study
The sampling frame consisted of nearly all women diagnosed with GDM in 2011 to 2012 across 44 KPNC medical facilities.
Here, we used stratified sampling within the GEM cohort to identify women from 6 ethnic groups representative of Northern California and among those with the highest prevalence and/or absolute frequencies of GDM.
Eligibility criteria included age 18 to 50 years; comfort reading and speaking English; not currently pregnant; absence of recognized overt diabetes, confirmed by the KPNC diabetes registry and self-report; and body mass index (BMI) of 25 to 40 kg/m2 among African American, Mexican American, and non-Hispanic white women, and 23 to 40 kg/m2 among Asian Indian, Chinese American, and Filipina women who are at greater risk for diabetes at a lower BMI.
Recruited participants (N = 35) had a mean age of 36 years (standard deviation [SD] = 5.3), and were a mean of 3.6 (SD = 0.3) years postpartum from their index GDM pregnancy. As designed, the sample was ethnically diverse (Table 1). Mean patient trust in the medical profession was 16.6 (SD = 3.8). Regarding perceived risk, 17% (n = 6) believed they had a high chance of developing diabetes, 29% (n = 10) a moderate chance, 49% (n = 17) a slight chance, and 6% (n = 2) almost no chance. Mean personal control over developing diabetes was 3.4 (SD = 0.5). Whereas 6% (n = 2) reported participating in a health system-based lifestyle program in the last 6 months, many endorsed intentions for the next 6 months with 69% (n = 24) somewhat or very likely, and 31% (n = 11) somewhat or very unlikely to participate. There were no significant differences across focus groups in any of the above domains (P ≥.07). Of 17 women in facilities assigned to the GEM intervention, 88% (n = 15) had participated in at least 1 session; participation was 50% among Filipina women (n = 2/4) and 100% within all other groups.
4. Limitations of the study
Study limitations include the sample’s relatively high level of education, their membership in a single health system, and, as noted, our limited ability to make cross-group comparisons.
Also, in my (reader’s) opinion, limited sample size and semi-qualitative approach.
-
1. Introduce yourself about your background, what kind of work you are doing that related to statistics.
I am a general practitioner (GP) physician in Nakorn Pathom, Thailand. I make and plan treatments to patients based on evidence-based medicine (EBM). I also observe patients for anecdotal data and personalized treatments, to confirm whether the treatments are successful or not.
As for the EBM, I have to rely on training courses, managed by various health organizations. I rarely truly use primary data for larger scale care planning, e.g. for the hospital / region, only for myself.
2. Have you ever learned or applied statistics in your work related to data analysis or statistical analysis. Please share your experience.
Not really, not for my career. I learnt to use statistics here (in BHI) to analyze COVID-19 more sensibly. But, I do hope to learn how to make better quality evidence-based data from my observation; that can better apply to my career. (For example, not some expensive medicine that I rarely use anyway; or caring more about drug compliance or ADR.)
-
-
Research Title: Interhospital data-sharing through patient: needs and concerns
1. What data have you, as a doctor, given patients to hold – categorical variable – summarized as proportion
2. Diseases, as a patient – categorical variable- summarized as proportion
3. What data do you as a patient or caregiver, bring to the hospital – categorical variable- summarized as proportion
4. Age (of doctors, patients) – continuous variable – summarized as mean (SD) and histogram
5. Occupation of patients – categorical variable – summarized as proportion -
-
1.
In my eyes, it might not be totally equivalent to blockchains, but important things are two things.
– Distributed system and decentralization
– Security, encryption, and authorizationBlockchains are more about decentralized banking, and perhaps Read-Write only databases, without Update or Delete, IMO.
2.
Because of recent news of ransomware in a hospital in Thailand, I think decentralization and security is important. Backing up of data is helpful, with a tint of DevSecOps, that is, security automation.
Therefore, strike iron when it is hot. While people are still aware, implement the system now!
-
1.
About COVID-19 contact privacy, although I might not be a good representative of general public (in Thailand), I feel that problems are two folds.
– Infringement on privacy. Though, it is mostly over-feared, when looked back.
– Hard to keep up consistency on picking up mobile phone app.Google Map fixed these well, when looked back into. Location services are automatic. It tells when there is a missing data. But yes, it is creepy to know that it knows where I have gone.
Lately, I noticed that there is a safer and more privacy friendly way with Thai Chana. You don’t need the app. You can just scan the QR code with LINE, and check in (and check out) with a web browser.
Also, as I know some programming, I do care what code is being run on the server or the app. It should be open sourced. Transparency is important to gain trust.
2.
Public view on academics is actually limited and comes from key influencers; so things can easily changed.
Yes, truthfully, I feel that academic researches can easily be leaked and valued for money. Can ethically committees, and IRB’s be trusted? I know good people tried their best to be good. But even at best intentions, good people can indeed be unknowingly manipulated, e.g. via selective evidence.
Still, R&D is the core of everything. When would it be best to take some risks, to ensure greater development and competitiveness, I wonder?
I cannot think like everyone, but I’d rather take risks, if potential massive benefits are promised.
-
1.
Biggest challenges are of two forms. Nature of big data itself, and systematic bias.
As for the first topic, which source of data should I use, at what time, and how to filter the data.
About systematic bias, one should consider the context of data makers; and not everyone are equal. For Twitter, there are key influencers, and is more likely to appear more on tweet timelines, therefore more retweets.
2.
Big data is more powerful, when the system of cleaning data, and making sense of them, is good enough. Yet, one has to be aware of biases, both systematic one, and randomness.
I would say that best sources of data, ethically, are via public API’s. As for Thailand, it might be via Facebook API, and to a lesser extent, Twitter. LINE might be another good source, if public API’s are present.
Another interesting source of data, that maybe ethically gray-zone, is web scraping. Also, web scraped data are much harder to make sense of.
-
-
-
Since we are using complete linkage clustering, the distance between “35” and every other item is the maximum of the distance between this item and 3 and this item and 5.
From the link, at the top of PDF can be found — Defining Cluster Distance: The Linkage Function
Complete linkage, which is more popular, takes the maximum distance.
-
1. There is a lot studies about creating a PHR prototype, why so few of them are implemented?
Whenever there needs to be a collaboration multiple parties (between hospitals, and between caregivers and general public), it is less likely to be implemented.
For example, PHR with hapi-fhir wouldn’t work without an HL7-compliant system, e.g. OpenEMS.
IMO, it might be better to start fragmented, instead of trying to do all at once. For example, PHR might be just some kind of specialized note-taking app at first. Don’t focus too much on HL7 (although it eventually should work with HL7).
Collaboration between hospitals in Thailand might not be some time soon; main issues being no trust in encryption / transfer of data between sites, as well as not allowing servers to be off-site or on cloud — fear that data will leak, or breached. Policy makers, e.g. hospital directors and MOPH, as well as government, have to get over this first.
Transferring data to PHR has an additional issue of data being out-of-hand of personnel (responsibility, training). Not trusting patients of always being an ally is another reason. (They can sue you, good will or not.)
2. In your opinion, what is the disadvantage of using PHR?
Mainly whether patients can be trusted with some health data, which can be complex and hard to understand.
Not only data can be interpreted the wrong way, depending only patients’ beliefs and untrained researching, but also patients might not understand hospitals’ contexts and limitations.
Of course, being electronic has issues I said in (1), that is — data breach.
-
1. Can human population movement apply to the other field of researching?
– Communicable diseases, obviously.
– There might also be issue with immigrated invasive species causing shifts in ecospheres, and possibly problem in living. Not sure how much it affects Non-Communicable disease and lifestyles, though.
– Traffic accidents / road accidents. Although it depends more on movement, rather than disease spreading / exposure.
– Psychology of immigrants / emigrants. Whether they can blend in, and impacts on communities they tried to live in.2. Is that OK to use the self-reported malaria to analyze? If no, what kind of data will you select?
There are issues with both self-reported and officially-diagnosed data.
Self-reported
– Earlier. Possible near real time (minus incubation period)
– More cases, therefore less likely to miss, but more false positives
– Some further analysis on association may help sometimes (like statistics, or tests for history-of-infection)
– Subjective bias is more possibleOfficially-diagnosed
– Always slower, as well as lapse time
– Less false positives
– Depends on whether patients seek official help in the first place, therefore bias.Is that OK? What kind of data will you select?
– I would say yes, but you have to be careful of whether it reflects the real data.
– It has the benefit of more cases, therefore cover more likely geographic area. Therefore, more alerts in the public health officials
– Both have be careful of lapse time, anyway. -
1. What are your opinion about electronic patient referral web application?
– It is fast, real-time, and relatively safe (no personal info leakage)
2. What is the problem about electronic patient referral web application?
– Digitization of useful data. It needs to cooperate well with the tools; otherwise, manual work is needed.
– Infrastructure is required. Although, it might be possible to adapt with low connection speed (e.g. telephone line/modem) and low CPU (e.g. older computers).
– Security and preventing failures all have costs, not to mention sys admins (human). Paper is usually safer. -
First of all, let me confirm that the paper is this link.
1. Why was the author interested in investigating the suicide problem in Thailand during the time?
– Suicide rate was increasing for a decade before 2011, and more than 6 / 100,000 in 2011, probably with a clear visualization?
2. Each of students picks one potential risk factor mentioned in the paper and explains how the variable can contribute to the suicide rate?
Not in Agricultural sector
– This is interesting, and maybe related to being hard-to-find-a-job and job/employer/workplace change. For agricultural sector, there is probably less likely a job change, despite not always generate much income — there is always more certainty.
3. How statistical modeling can contribute to investigate the epidemiology and spatial aspects of Thai suicide problem?
– It is a hypothetical problem, or a real problem?
– Association or causal relationship? Cause or effect?
– Is funding in the right place?
– General understanding of the context of that place or society. -
ERROR: dependencies ‘sp’, ‘foreach’, ‘shiny’ are not available for package ‘INLA’
* removing ‘/Library/Frameworks/R.framework/Versions/4.0/Resources/library/INLA’
Warning in install.packages :
installation of package ‘INLA’ had non-zero exit statusFor MacOS, you also need ‘shiny’.
So, install ‘sp’, ‘foreach’, ‘shiny’ first, then try install ‘INLA’.
-
Discussion 1
According to this paper, the preventive measure achievement rate is around 50-70% of patients. The implementation of CDS did not improve the rate of preventive measures as much as we think when we use the same calculation. If you were a CDS developer, which features or options you want to add on in order to improve the performance of this CDS?– Reminder system, calendar and notifications are definitely good ideas, as long as users doesn’t have too many of them and block them themselves. I personally have both too many emails and too many mobile push notifications.
– Silent data collection would be nice, but in reality, can only be done in limited ways.
– Personal data collections might be done in wearables, such as Apple Watch.
– I don’t know how much Anonymous data collection helps, but there are always privacy concerns.Discussion 2
If this study was conducted in Thailand, do you expecting the same result? Given we have a semi-open health care system. The universal health coverage program takes preventive measures as a priority. However, some preventive measures, such as vaccination or mammography or medication for osteoporosis treatment, cannot reimburse. Most general practitioners spend time around 5-10 minutes for each patient.– The result would be different by culture. In Thailand, patient generally rely on officials, such as doctors, too much, and know very little about what doctors give, such as medicine.
– Actually, currently, people that tried to learn can be very superstitious, and sometimes unreasonably resist the doctor. This is maybe because personal heath knowledge in Thailand is still in early stages.Most general practitioners spend time around 5-10 minutes for each patient.
– This is actually because doctors in public sectors are overworked, and leakage into specialities and private sectors exacerbates this.
– IMO, health organizations should advise patients to prepare more before visiting the doctor, as well as holding on to Personal Health Record (PMR) themselves. Disease prevention advice can be generated programmatically, perhaps also with AI. -
Topics 1
Good electronic system come with price , how do we develop it in low resource country like Myanmar?I think we have to separate things.
Infrastructures, such as fast Internet are expensive, and have no other to replace. A solution is to work offline-first, and sync only when you have Internet connection, such as EpiCollect5. Web development trend of Progressive Web App (PWA) is also on this concept, but I don’t think it is there yet. We are moving from always online state of web, to can-be-offline state of mobile apps.
Hardwares can also be expensive, but they are movable, therefore easy and quick to migrate from richer areas to poorer areas, and that highly depends on government (and NGO) funding.
Softwares are not expensive per se, but they are difficult to both develop and maintain, as well as can easily be brittle and buggy, and always need support. We need localized smart people. We need scholarships to local people to help keep them to maintain the softwares.
Topics 2
How do we manage the capacity of staffs with high turn over rate?There are two ways to correct this,
– Optimistic way is a little difficult, but something can be said
– Use more local and loyal people
– Give people sense of security
– Give non-monetary rewards, such as welfare???
– Better the infrastructure???– Pessimistic way is, Rely less on skills and in house training, but rely more on stronger systems. The rest is newer staffs just have to “fit in”. The only takeaway here is, stronger system, though.
-
1. What is your opinion about Concentrations of SO2 in the resident more than workplace .
It is hard to guess, but I would consider the following factors,
– Windswept
– Non-industrial activities
– Household activities2. What do you think about this paper to improve paper adapt to Thailand.
A research paper should be accompanied by well-thought dashboards and infographics for decision makers; that is, not only fill by region-specific data; but also serve well for shorter-than-ever attention spans of stakeholders and decision makers. It would help to think that they don’t always have much time to listen.
It has to both be region specific, such as Northern Thailand; and race with other topics for attention, and catalyzing changes. As always, higher ups can both have too many things to improve, and sometimes might even be against their interests.
-
Please see my COVID19 Dashboard.
I feel PowerBI a little slow and lagging for my wanting to make decisions…
-
Map is nice, but if we try to make it a flat projection for the whole world, it is distorted.
Funnel chart is just akin to Bar chart.
Tree map is a very good idea to compare numbers.
Scatter chart is a good idea to show relationship between 3-4 variables, but I personally think time series animation is too complicated and not always understood immediately.
-
– Sparkline is used to show general trend of Confirmed case over time.
– Line chart with forecast is used to show future trend, with confidence interval.
– Clustered bar chart makes it easy to compare between categories.
– Donut chart and sunburst makes a nice comparison over 100%. -
Please see my dashboard.
Table vs Matrix are different kinds of comparison, but Matrix can easily more powerful and get what I want; but it can easily more complex. Maybe I don’t yet know how to do much with PowerBI transform data…
IMO, I am more interested in ratios, probably to populations, as it can be more nicely compared.
-
Everyone seem to use Mercator projection map for some reasons, even though it is equator-small, Greenland stretched, and USA-just-the-right-size. I might even propose something like Google Earth or D3.js.
– https://covidvisualizer.com/ (This seems to be quite nice and delivers right information at the right location.)
– https://covid3d.live/Also, my Chromium-based browser, Brave, lags because of some maps… Not sure if it is the case for Firefox and Edge, which use different rendering engines.
I personally like the idea of not-too-long a table.
I have just found this Brazil-focused map, and it uses interesting design of Donut charts. Bar chart is also very helpful in this case, although it might be bettered by sort frequency by DESCENDING or something. https://sigageomarketing.com.br/coronavirus/
-
Dengue and Arboviral diseases are nice diseases to be studied and apply for Thailand, especially coupled with mapping system — as I saw that there is much affected in southern of Thailand.
As for the current state of Thailand, when the disease is reported, mosquitoes must be controlled, so yes, GIS is important.
-
> Do you know there is any data-driven decision-making facilitation using in Thailand? If not, which public health issue you will start with?
Sadly, I don’t know one, but data-driven organization is a direction.
> if you were a researcher, what step/detail do you want to add or change in this process flow? and from research limitation, what factor do you think that can adjust for the better study, for example, population (a small number of interviews), stakeholder management, is there any else in your opinion?
Easiest to do, is probably to decentralize interviews, but it is hard to control the quality. Make it a checkbox, or use Google Forms is probably good, but pilot interview is needed first. That’s why this kind of study is needed.
-
-
-
-
-
If there is not yet full paper available for the second one, it is hard to choose. But for now, I can tell that for the first one, you might be able to learn a lot about GIS (compared to some other papers I have seen).
– Quite new, about five years
– Time series
– Spatial mapping
– High impact disease -
– SubjectID can be backtracked to confirm site of working; also included are date of visit and date of informed consent
– Exact date of birth and initials should probably be avoid, as it can potentially identify the person. I would rather reduce to Month and Year, or Year only, just to be safe.
– Onset date in General Medical History is picky, and consentee may have multiple problems per system.
– Otherwise, I think it is good. -
A direct and clear advantage, is that when that research is picked from the shelf, it can be correctly interpreted. Meta-analysis can be easier.
In the meantime, before the research is done, it can mean that it can be computerized and OCR’d, and data sharing can be easier to be done in real time. Also, if there are multiple implementors, the data can be accumulated into one.
-
I have seen a study of a pharmaceutical that we have to sign who is interpreting the questionnaire, but the use simply Excel.
Still, in many parts, it is excruciating and error-prone. If SPSS is to be used it might be safer. A relational database isn’t used yet. Using simply Excel, access control isn’t there yet.
-
-
In my experience of data collection, I have little experience of research, but the data collection system has to be well designed. However, I truly missed one facet, what if the data is reused…
I can tell little from the point of view of software development and utilizing data. When a long time has passed, maybe months, I tend to forget how the data works, and the data can be reused only if,
– It is plain text, and can be checked by eyes, although you might automate a little. This also includes CSV and TSV.
– It is a database with well defined schema. SQL is easier to be reused than NoSQL (unless the API controlling the NoSQL is able to be read, i.e. clean code.)Also, if you have a well documented API, probably via OpenAPI or Swagger, it is easier to be reused (I think).
I only have a real collaboration with multitude of unknown people in one instance, Version Control System (VCS), namely GitHub. Every steps have to be traceable, and by whom. Also, it is best to set guidelines and style guides. Also, the commit message telling why it is updated should adhere to a guideline as well.
On a real database, I have used databases that have user system, such as PostGRESQL and MongoDB. When I let user to edit the database (i.e. CRUD), such as a questionnaire or a commenting system, I have to be extra careful on security and hacking…
There is a common saying in programming — never trust user inputs. We have to always sanitize the input in every layers, including the internal ones.
-
I was indirectly involved in data extraction from questionnaires of my parent’s work (of a pharmaceutical company).
– It might be impossible with design the questionnaire without marketing research.
– Other checkboxes, and fill in the blank might the choice, but the answers vary; and it is prone to subjectiveness of what is similar and not.
– Further comments is also subjective. We should rely on better designing a questionnaire.As a questionnaire answerer, I find that too long a questionnaire means that the result is unreliable. Also, analog scale of good to best (1-5) is very subject of what is 3, 4, or 5.
-
I think we should find and train more people who could create values in Health Informatics, leading a project, and creating a health product. Only then can we count competent personnel in Health Informatics.
Distribution is equally important as well, just like in PMAC clip about UHC, where a smart local people works devotedly to help their friends in remote area. In my experience, it doesn’t have to be local people, though; but we just need a devoted people, who are determined to work 4-hours distance far away…
-
Thank you very much for sharing. It is always nice to see more code.
However, this makes it confusing with Local Moran’s I
– A positive value for Ii indicates that the unit is surrounded by units with similar values.
– The Getis-Ord Gi Statistic looks at neighbours within a defined proximity to identify where either high or low values cluster spatially.The page also mentions Geary’s C which also found in Wikipedia some time ago.
Not sure if exceedance probability is considered a spatial statistics…
EDIT: The rpubs links to a lot of YouTube VDOs, which is very helpful 🙂
-
Most updated lab results and vital signs seem to have no substitute, therefore should be easy and best to implement first.
OCR using image technology might be good, if it is accurate. Forget handwriting, it has to be prove-read first. Older printings and PDF might be worth OCR’d — but it is still better to just upload the doc files…
-
