Presentation is loading. Please wait.

Presentation is loading. Please wait.

Medical Informatics Workshop 2003-2013 Prognostic Matches and Mismatches C(4) - HIS: Research and reporting (T22 - T23) Victor Maojo and Frank Ückert.

Similar presentations


Presentation on theme: "Medical Informatics Workshop 2003-2013 Prognostic Matches and Mismatches C(4) - HIS: Research and reporting (T22 - T23) Victor Maojo and Frank Ückert."— Presentation transcript:

1 Medical Informatics Workshop 2003-2013 Prognostic Matches and Mismatches C(4) - HIS: Research and reporting (T22 - T23) Victor Maojo and Frank Ückert

2 Introduction In general, it is quite dangerous to make accurate predictions in Science and succeed in such process. Examples: – Herbert Simon (1956): “In 25 years computers will make the same intelligent tasks than humans do”. And he lived until 2001… – Simon Newcomb: “We can never build machines that fly”, he said (and justified mathematically), shortly before the Wright brothers made the first plane – A consultant to Eckert and Mauchly, creators of ENIAC, the first electronic digital programmable computer: “the world market for computer sales is a maximum of 6 units”

3 T22: The type and extent of health care reports will continually increase. It will be possible to publicly call up and evaluate aggregated data of in and out patient diagnoses (e.g. current prognoses in regard to ‘health observation centers': announcement of epidemics and infectious diseases, etc.). ++ Healthcare reports continuously increase, given the extension of social networks and people (not only health professionals) that can contribute by accessing information +- The idea of “health observation centers” has not succeed as such (this one is the only mention in Google), but it is interesting to see how many people (even individuals) can develop interesting analyses. They can become some kind of “personal analysts”, by using personal intercommunication combined with Google searches and information from different sources (web, literature, documents, social networks such as Patients like me, Twitter and Facebook, etc). For instance, using Google to detect the spread of diseases or using social networks to find non previously known secondary effects of concrete drugs + Automatic reports not yet available, but text mining tools help to develop new ways of extracting and managing information. Problem: many languages, lack of corpora for each area, different records and standards, lack of dictionaries, semantic issues. This is still a manual (actually, semiautomatic) process

4 P22.1: In and out patient diagnoses are available via Internet in the form of aggregated data. ++ Statistical and literature data can be aggregated to EHRs (e.g., infobuttons, plugins that associate links and literature searches to particular words in a EHR, etc) ++ Aggregated data include multilevel (-omics, organ, cellular, public health, clinical data) and multimedia (X-Rays, drawings, videos, Web captures, etc) information --!!! Threats: comments by non-professionals (personal mails, forums, mailing lists, non-professional websites, fraud companies which sell fake drugs, non-certified doctors) can make people believe to have imaginary diseases or understimate real threats in order to make business

5 T22 (aggregated diagnoses online available) in 2013? - No real online tool for simple, free and on-time access of clinical data available. Three examples (no claim for completeness!) of similar, imperfect services in Germany on following slides:

6 Reportable diseases from the RKI + Possible to retrieve data over the internet coming from clinicians on the basis of the law ”Infektionsschutzgesetz (IfSG)“. + Special tool called “SurvStat@RKI“ gives possibility to query the data freely and individually, resulting in own tables and graphs.

7 “Gesundheitsberichterstattung (GBE) des Bundes“ [health reporting of the state] + Free of charge, more than 2 billions of numbers regarding health in form of tables from more than 100 different sources. (CAVE: not all diseases) + Aggregated diagnosis data (ICD 10 GM) from year 2000 online accessible, but max to three positions (level of “Allgemeine Systematik”). As example: “A95: Gelbfieber” + On this level there are also other analysis available. As example: “Kinderkrebsregister” or statistics of death reasons (Todesursachenstatistik) from the “Statistisches Bundesamt”.

8 Getting close: DIMDI‘s “Informationssystem Datentransparenz“ Usage of existing pseudonymised data reports from health insurance companies to the ”Bundesversicherungsamt für den morbiditätsorientierten Risikostrukturausgleich (Morbi-RSA)”. – [sorry, that one we have to explain non-germans in person. I am sure there is no fitting translation…] Aggregation of this data over several years with a second level pseudonym. Basis for data aggregation: German law named ”GKV- Versorgungsstrukturgesetz“, came into effect at 01.01.2012. Basis for usage: §§303a bis 303e of German law named ”Sozialgesetzbuch V“. CAVE: One has to write a proposal first, but for scientists it is explicitly allowed. First data will be available in the second half of 2013. Kind of medical data: Information on insurance company and insured, diagnoses (ICD 10 GM), prescriptions (”Pharmazentralnummer“), procedures (OPS).

9 Other countries NHS initiative CPRD: claim to have all clinical observational data for (registered) researchers – Kind of service? – Raw data? – Claim to use several classifications (ICD, SNOMED, …). Translation?

10 P22.2: Less than 5% of the diagnoses will be up-dated on a weekly basis. -- Almost impossible to measure such issue with a universal perspective. Differences tend to diminish in many topics (e.g., use of EHRs, radiological devices, etc) but increase in others (e.g., due to economic reasons in this time of crisis) ?? EHRs, main problem in MI. Still companies reluctant to adopt standards due to economic reasons

11 T23 (clinical research on online data) in 2013? First real online tools for simple, free and on-time access of clinical data available at some university hospitals, but still of a project character, only for pieces of data (like for a group of diseases, one of several clinics/departments or one network of researchers) and only partially embedded in clinical routines. Usually the data is gathered (extracted manually or automatically after manually providing an interface, then copied) only with a formal reason, like a clinical trial. Therefore specialized clinical trial software (e.g. macro, openclinica, secutrial) or “self made” registry software is used. Trends for registries in oncology (e.g. for certification reasons and quality management) as well as in large style biobanking, pushes the development of those online available clinical data for research. Two examples (no claim for completeness!) of still imperfect, but existing, services/projects in Germany on following slides:

12 HIS-based support for the recruitment of patients for clinical studies (BMBF KIS Rek) Patient recruitment can be supported electronically by using patient data from routine documentation. E.g. study assistants may be notified automatically. So far only some university hospitals perform that within their HIS. A generalized implementation was the work of “KIS Rek”. [Rainer might know much more, because he is part of the team.] Same scope in other, larger EU-projects, e.g. EHR4CR (Innovative Medicines Initiative) or PONTE (FP7).

13 DataWarehouses I2B2 in Erlangen: i2b2 (Informatics for Integrating Biology and the Bedside) is an NIH-funded National Center for Biomedical Computing based at Partners HealthCare System. The i2b2 Center is developing a scalable informatics framework that will enable clinical researchers to use existing clinical data for discovery research and, when combined with IRB- approved genomic data, facilitate the design of targeted therapies for individual patients with diseases having genetic origins. In Heidelberg: Complex connection in series of THREE data warehouses to prepare data from routine systems for research. Partly finished. Some (few!) more…

14 T23: Clinical research will be supported by online data, however, only if electronic patient records are available and the data has been stored in a structured manner. It will become evident that comprehensive, systematic planning will be necessary to make use of the data for research. + EHRs more difficult than expected. “A unique, universal EHR will be ready in ten years” (something that was more or less expected in1993, then 2003, then 2013…2023?) + Semantic interoperability: going slowly ++ Large plans can overcome by individual and small groups and companies with rapid initiatives, with developers and users targeting users with personal tablets and mobile phones

15 There are still many places with no EHRs…

16 T23.1: Hospitals storing over 80% of all patient documents electronically (compare P14.1) will base over 50% of their clinical research publications on online data obtained in routine clinical tasks. EHRs, main problem in Medical Informatics. Companies still reluctant to adopt standards due to economic and strategic reasons Patients as “customers”, companies use data from these “Personal Health Records (Mandl and Kohane, NEJM) Plugins, connecting to the literature (as explained above)

17 The theses of 2002 T22: […] publicly call up and evaluate aggregated data of […] diagnoses – P22.1: […] diagnoses available via internet […] – P22.2: Less than 5% of the diagnoses […] updated on a weekly basis T23: Clinical research will be supported by online data, […] comprehensive, systematic planning [for storing the data in a structured manner first] will be necessary to make use of the data for research. – P23.1: Hospitals storing over 80% of all patient documents electronically will base over 50% of their clinical research publications on online data obtained in routine clinical tasks.

18 Final Congratulations to Reinhold and coauthors by making statements that were of great interest (and challenging) ten years before


Download ppt "Medical Informatics Workshop 2003-2013 Prognostic Matches and Mismatches C(4) - HIS: Research and reporting (T22 - T23) Victor Maojo and Frank Ückert."

Similar presentations


Ads by Google