Background: Generally, in a society working with efficiency and budget requirements (i.e., limited resources), the aim should be to provide maximal effectiveness efficiently with the given inputs (technical efficiency) based on optimal decision criteria (allocative efficiency). Health economics and outcomes research (HEOR) can also pursue effectiveness (maximize health gain) and equity (minimize health inequalities) – some aims of the Health and wellbeing, a Finnish governmental key project . In larger scope, health technology assessment (HTA) pursuit e.g. efficiency and equity with clinic, HEOR, ethic, organizational, social and juridical evaluations.
Efficiency and equity should be pursued based on solid and up-to-date HTA evidence. As data and evidence accumulate over time, the HTA content should be updatable, real time and predictive. In eHealth HEOR (e.g. Virtual Hospital 2.0), a use of 1) existing data, 2) predictive analysis, 3) managed knowledge, 4) targeted communication, 5) multi-stakeholder implementation, 6) outcomes assessment, 7) efficiency and equity evaluation, and 8) active corrections was considered relevant [e.g. 2, 3]. Namely, eHealth should be available for all in need, real time and customer responsive . Such eHealth may free considerable monetary potential [2, 3].
Unfortunately, trade-offs between efficiency and equity  as well as between efficiency and affordability are common . All of these increase the need for accurate and timely individual-level data. Data lake and service operator projects (Isaacus [e.g. 6-8]) aimed at providing such data fast and securely for various purposes.
Aim: To describe some potential uses of individual-level data lake data (DLD) from HEOR perspective.
Results: HTA should consider multiple viewpoints and be logically established and reported. An approach suited for the HEOR part of HTA requirements is PICOSTEPS principle, which was built as a supplemental part of Finnish Current Care Criteria  and is used in e.g. real-world data based  and modelled  HEOR.
DLD-associated systems such as remote use platforms or encrypted delivery of DLD have the potential to provide fast and secure access to data readily compiled in a single place. In order to be of use, data and metadata need to be clear. Time-series can be available in DLD, which is very important in establishing e.g. characteristics and natural progression of individuals, service pathways and outcomes over time. With such data, the common challenges of HEOR – who is cared for, how, when and what are the consequences over time – can be addressed.
Examples of HEOR effects that can be potentially estimated from DLD include clinical effectiveness, safety, surrogates, resource use, costs, quality of life and survival. Examples of effect modifiers potentially established from DLD by using statistical or machine learning methods include age, sex, conditions, comorbidity, genetics, sequences of treatments, family, care provider at different organizational levels and care pathways.
By using predictive analysis, DLD can be used to establish expected effects and condition/disease progression patterns for HEOR assessment tools such as cost-effectiveness , cost benefit [3, 4] or budget impact  models commonly applied in e.g. funding decisions or recommendations. Other examples of DLD use include profiling and care targeting. These can be used for the knowledge-based decision making. Additionally, DLD could be applied for outcomes-based implementation such as risk-sharing – a kind of a warranty for health technology. Risk-sharing has the potential to simultaneously gain health and economic benefits [13, 14].
In HTA, targeted, appealing communication based on established methods is important for result applicability and implementation. In the best case, all stakeholders (e.g. patients, sectors, government) are involved [e.g. 15].
Conclusions: DLD can be used efficiently and securely to meet many HEOR needs. Considering technology-related opportunity costs [e.g. 16], eHealth potential [e.g. 2, 3] and practicalities [e.g. 14], DLD together with suitable analytics and eHealth is a potential treasure chest for HTA. However, care should be taken to develop legislation and infrastructure to facilitate the optimal use of DLD and other data sets for various purposes.
References:  valtioneuvosto.fi/en/implementation-of-the-government-programme  Virtual hospital 2.0 – modelled cost-benefit assessment. eHealth 2018  Predicted cost-benefit of Virtual Hospital 2.0. WHO Healthy Cities 2018
 Soc Sci Med 2018;212:136-44  benthamopen.com/contents/pdf/TOALTMEDJ/TOALTMEDJ-4-1.pdf
 sitra.fi/en/projects/isaacus-pre-production-projects/  Value Health 2017;20:A777  Value Health 2018;21:S217
 kaypahoito.fi/web/kh/suositukset/suositus?id=nix02465&suositusid=hoi50062  Clin Ther 2017;39:537-57.e10
 ClinicoEcon Outcomes Res 2018;10:279-92  ESC Heart Fail 2017;4:274-81  Adv Ther 2017;34:2316-32
 Terveystaloustiede 2012. THL, 69-73