The GetSet’s Post

Discover how online Master's programs in Biostatistics combine statistics and healthcare, offering accessible education and diverse career opportunities—even for those without a related undergraduate degree. 👇Read more https://lnkd.in/eDk9Bj26 #biostatistics #statistics #data #science #education #career #blog --- At biostatistics.ca our mission is to create an information-rich community around biostatistics. Do you have interesting career stories, project successes or technical skills you would like to share with the community? Send us a message and become a certified biostatistics.ca writer!

Accurate measurements and data analyses are invaluable throughout healthcare systems. That is, sound statistical methods in clinical trials, university research, and hospital operations are crucial to better patient outcomes. Medical statisticians – also known as biostatisticians – are key to setting the foundations of medical research and to clinical decision-making. Dive deeper into this data-driven career in the most recent Global Campus blog. #statistics #StatisticalAnalysis #statisticsclass #statisticsmatter #biomedical #BiomedicalScience #biomedicalresearch #biomedicalengineering #biomed #DataScience #dataanalytics #DataDriven #healthcare #healthcarecareers #medicalresearch #medicalstatistics #Biostatistics #biostatistician Adam Wellstead Rick Berkey Brian Hannon Jacque Smith Melissa Keranen KUI ZHANG John Gruver Michigan Technological University Michigan Tech Career Services Michigan Tech College of Computing Michigan Tech College of Sciences and Arts https://lnkd.in/gc8zCeME

Confounding can distort the true relationship between exposures and outcomes in observational studies, making it crucial to identify and account for potential confounders to avoid biased conclusions. In a new collaborator article, Jesca Birungi shares a great introduction to this topic, along with some examples. 👇Click below to read more https://lnkd.in/eZTu5CxN #biostatistics #statistics #data #science #education #career #blog — At biostatistics.ca our mission is to create an information-rich community around biostatistics. We're currently looking for 1️⃣ PARTNERSHIPS to help our community 🤝 Brand partners, companies, or startups that are developing interesting products for our audience. 💯 Admission Experts who could help students get admitted to the best program for them. 🦉Career Counselors who could help people navigate change and opportunity in their careers. 🎯Headhunters and Interview Prep Specialists who could help people land their dream job. 2️⃣ COLLABORATORS - WRITERS to share with the community  Do you have interesting career stories, project successes or technical skills you would like to share with the community? Send us a message and become a certified biostatistics.ca writer! Please, don't hesitate to reach out via DM ✍

In our brand new collaborator article, Eric J. Daza, DrPH, MPS explores the evolution of his work from causal inference in biostatistics to pioneering methods for observational causal inference using n-of-1 time series data, driven by the promise of personalized digital health insights. 👇Click below to read more https://lnkd.in/eiiMQaWk #biostatistics #statistics #data #science #education #career #blog #causalInference — At biostatistics.ca our mission is to create an information-rich community around biostatistics. Do you have interesting career stories, project successes or technical skills you would like to share with the community? Send us a message and become a certified biostatistics.ca writer!

Junrui Di, who received his PhD from the Johns Hopkins Department of Biostatistics in 2019, is a Digital Medicine Statistician at Pfizer. Read about Junrui's experience in our PhD program, his current work as part of Pfizer Research & Development's AI/ML Quantitative and Digital Sciences group, and his advice to current and future students, below. Junrui writes, "It's indeed cool when the quantitative knowledge and skills we possess can directly benefit people's everyday lives, making a difference that matters."

Thanks to Jesca Birungi for contributing this great article to biostatistics[.]ca on Causal Inference, more specifically confounding in observational studies. Check it out 👇

Biostatistics

Biostatistics 101: The two schools of statistical inference – Frequentist vs Bayesian. A 2-minute overview of the two and how they differ. Within statistics, there are two primary methodologies. Both are useful for data analysis as long as they are interpreted correctly. Frequentist – is the main school of statistical thought used in medical literature and clinical trials and is based on null hypothesis significance testing (p-values and confidence intervals). Bayesian – less common in life sciences but still used and is based on prior beliefs to develop new “posterior” beliefs.   The key difference between these two schools is how they interpret probability and uncertainty. The frequentist approach assigns probability to data (how likely were we to generate the sample we did from the population) while the Bayesian approach assigns probabilities to hypotheses (how likely is our hypotheses given the observed data and prior information we have). This makes the Bayesian approach MUCH more intuitive to interpret. How likely is my hypothesis is a lot more straightforward to understand than “how likely was I to observe this sample given the null hypothesis is true”. So it begs the question: “Why does the vast majority of life sciences research use the frequentist method when it is so easy to misinterpret the results compared to a Bayesian approach?”. There are a couple of reasons – but one of the main reasons is due to how Bayesian statistics works. Bayesian statistics uses prior information in addition to the data sampled to derive inferences and to produce probabilities around the hypotheses. This is great, however, it can be very difficult to find accurate and reliable prior information. More information on how we identify and use this prior information coming in future posts. Follow me and Better Biostatistics for more. Happy Monday 

Impute the missing data or not? - one has to think twice, especially in Biostatistics. While many would consider a priori defining which is the best missing data strategy, first, its wiser to answer the question - is imputation needed or not? An if needed define the best strategy and methods. Estimands are very important here - a Randomized controlled trial with treatment policy estimand framework, missing data imputation is generally almost a must. Not imputing the data in this case would break the whole principle of the estimand and potentially the randomization too. Interpretation potential, study design, known and unknown covariate bias correction power of randomization could all be lost to a certain degree. For this scenario a good imputation method is a very important, multiple imputation/machine learning based are my number one choice, but again depending on the explanatory power and predictive accuracy of covariates, many moving parts need to be considered and sometimes other methods may be more useful. If the design of the study does not require missing data imputation and bias from the missing data can be avoided, sometimes its actually much better not to impute the data. No way of imputing the data can replace the real data, so if the estimands allows it, the method implementation allows it and the bias from not having a segment of the data avoided, the best case is not to use imputation at all and keep the actual 100% original data. For this second scenario, an information theory perspective must be taken. No new information can be created in the dataset by any time of missing data imputation, but small amounts of new bias can be introduced. So in the second scenario its better to stay within the original data, but under the condition that no other biases such as survivorship bias are not introduced. This question may also depend on the method. In PCA we might need to impute the data as example, but we also need to consider the amount of missing data and if a better option is to exclude listwise. Listwise exclusion might introduce a variety of biases, but so may the imputation. Balance between the amount of missing data, in terms of potential deletion or imputation is very important here. Another very important question is the covariate predictive accuracy, in case of a high predictive accuracy and explanatory potential of the predictors used to predict the missing data, bias introduction risk will be smaller and larger amount of data can be imputed as a result (not too large in any case). With smaller predictive accuracy and explanatory potential of the predictors used to predict the missing data, no large amount of missing data should be imputed, at least according to my recommendation (there are exceptions). This area requires Biostatistical, domain specific, data science and study design knowledge / requires a lot of skills experience. #biostatistics #datascience #research #analytics #clinicaltrials #rct #lifescience

🩺 Essential Guidelines for Reporting Statistics in Medical Research 📊 🌟 Accurate and transparent reporting of statistics is critical for the reliability of medical research. Here are the key guidelines to ensure robust and clear statistical reporting: 1️⃣ Follow Existing Reporting Guidelines: Utilize specific guidelines for different study types: #CONSORT for randomized trials #STROBE for observational studies #TRIPOD for prediction models #AMSTAR for systematic reviews 2️⃣ Detail Statistical Methods: ✔ Describe the study questions and the statistical approaches used for each. ✔ Ensure methods are detailed enough to allow replication by an independent statistician with the same data set. 3️⃣ Proper Use of Inference and p-Values: 🚫 Avoid accepting the null hypothesis outright. 📉 Recognize that p-values just above 0.05 are not indicative of a trend. 🎯 Understand that p-values and 95% confidence intervals do not quantify the probability of a hypothesis. 🧩 Do not use confidence intervals to test hypotheses. ♀ Be cautious when interpreting multiple p-values and avoid separate p-values for different groups to infer differences. ♀ Use interaction terms instead of subgroup analyses and avoid uninteresting tests for change over time. 4️⃣  Reporting Study Estimates 🔸 Use appropriate precision levels, avoiding overly precise numbers. 🔸 Prefer medians and quartiles over means and standard deviations for descriptive statistics; avoid ranges. 🔸 Report estimates and confidence intervals for main study questions. 🔸 Treat categorical variables and continuous variables appropriately; avoid unnecessary categorization. 🔸 In meta-analyses, do not ignore significant heterogeneity. 🔸 For time-to-event variables, report events and median follow-up for patients without the event, not just proportions. 5️⃣ Ensure Clarity and Precision: ✔ Report p-values to a single significant figure, unless close to 0.05, then use two significant figures. ✔ Avoid reporting p-values as "NS" for non-significant results; use precise values like <0.001 where appropriate. ✔ Avoid reporting redundant statistics; ensure each number provides unique information. 📝 Conclusion: 🔵 Incorporating these essential guidelines into your research practices will not only improve the clarity and precision of your statistical reporting but also strengthen the overall quality and credibility of medical research. For more information, you can refer to 🔷 https://lnkd.in/dFXmkekd 🙏 Thank you all for visiting and engaging with my post. Your support means a lot! 🌺 ☘ If you found the content valuable, I would greatly appreciate it if you could share it with your network to help spread awareness. 🙏🦋 #MedicalResearch #StatisticalAnalysis #DataAnalysis #Manuscript #Article #Paper #Research