I don’t find where I can collect data or upload data, and how a cohort is generated after cohort definition.
Hi, @DreamCatcherlx,
I think you’re comming from the perspective of a prospective researcher, where you define your study parameters, define your protocol for data collection and move forward in time collecting your data for research, and performing analysis.
While the CDM technically could be used for this purpose (it has all the requisite fields to store patient level information, and a very verbose vocabulary to identify many different kinds of events (drugs, procedures, diagnosis, or general obsrvations), the CDM is a standardized datastructure allowing different data providers to migrate their data into a standard format and then we can perform large scale analytics on it. In this mode, the focus is on retrospective studies.
But if you’re working in the prospective mode, it is possible to standardize your data collection into a CDM form, and then at the end of your data collection, you treat the data ‘retrospectively’ and use the tools to identify populations via cohort definition, explore the data collected via Achilles, and do comparative cohort analysis via CohortMethod.
If you are asking for a description of how a researcher would use Atlas from a retrospective perspective, I can give details on that, but I thought you were asking about collecting and uploading data into a CDM from your question, so I’ll stop here.
-Chris
Firstly,thank you for your detailed explanation. But, I still have some problems:
I know ATLAS is intended for cohort data analysis, and I would like to know if there are any requirements for source database.
From your description, can I think ATLAS can just standardize data collection to CDM form? I would like to confirm if ATLAS provides cohort data management?
As we all know, now many popular statistical analysis software,such as SPSS, SAS etc, serve for cohory study data analysis, I believe your ATLAS has some advantages over common analysis software, can you explain the need for me?
I’m not quite sure I understand the question, but I don’t think that there’s any constraints on the source database, it’s only a matter of translating what the source system represents into how the CDM wants to represent the information. There may be gaps, but the CDM structure is primarily focused on the patent care, vs something like hospital staff management (it’s not an ERP system, it’s a EHR at the patient level).
Atlas isn’t for data collection. It’s for OMOP CDM Vocabulary exploration, chort definition and study design (as in the methods used to apply against a datasource to produce clinical evidence to support clinical decision making (ie: what’s the best treatment for patient X given medical history events Y).
If you ask a dozen SAS programmers to write an analytical procedure, they will each write their own code. From an OHDSI tool perspective, we’re trying to take the guesswork out of those decisions and produce consistent, repeatable analyses that can be shared across a network of research nodes, without making those nodes share patient level data between them. Instead, the instructions to execute a series of analyses (what we’d call a ‘study’) is distributed to each node, and the nodes have the decision to either execute and share results, or not engage at all.
We’re also developing the ‘best of breed’ visualizations for representing the results of a study visually. I use the term ‘best of breed’ because we have many different perspectives on how a particular set of statistics should be visualized, and the objective is that the most suitable representation that comes from the community becomes a standard package offering within the OHSDI tools.
-Chris