The components of the LAMP Platform rely on IT infrastructure that can be self-hosted by your organization (sometimes called "on-prem") or hosted by a cloud provider such as Amazon or Google.
If you'd like to deploy your organization's instance of the LAMP Platform on a cloud provider, please consult with your legal and IT departments first, and always ensure that you have signed a Business Associate Agreement (BAA) with the cloud provider to comply with HIPAA.
There are a few factors that need to be accounted when determining how to deploy the components explained above, as well as how high your monthly costs may be.
Computing: The components of the LAMP Platform are highly efficient and won't need much computing power; however, at minimum they require a single dual-core node with at least 2 gigabytes of memory.
Storage: The data collection rate for the LAMP Platform varies significantly depending on usage needs, spanning from a minimum of 250 gigabytes of low-throughput (HDD) storage to a recommended 4 terabytes of high-throughput (SSD) storage.
If you are collecting digital phenotyping data, using the integrated development environments, or scheduling intervention delivery/data analysis scripts, the minimum requirements WILL NOT be sufficient.
Network: The network bandwidth of the LAMP Platform varies significantly depending on your organization's size and study/clinic requirements; at minimum you must have single virtual private endpoint (i.e. firewall) with a bandwidth of at least 1 Gbps, to a recommended 10 Gbps to adequately handle sustained multi-user loads.
In all cases, the most costly and intensive part of deploying and maintaining the LAMP Platform will be storing the data.
When hosting the LAMP Platform with a cloud provider, your monthly costs may vary significantly based on the requirements explained previously. Here are two example configurations with the Amazon Web Services (AWS) cloud provider:
- "I run a digital clinic with about ~15 patients and don't intend to use advanced features such as digital phenotyping (sensor data) collection or Automations."
- Computing: 1 node, 2 vCPU, 2 GB RAM
- Storage: 250 GB, no snapshots
- Network: Up to 10 Gbps
- Total Cost: ~$35.00/mo
- "I run multiple research studies with 200+ patients and intend to collect digital phenotyping data at a high frequency, develop and use machine learning algorithms through the Automations framework, and more advanced features."
- Computing: 2 nodes, 16 vCPU, 32 GB RAM
- Storage: 8.0 TB, automated monthly snapshots
- Network: 25 Gbps
- Total Cost: ~$1,500.00/mo
Please note that it is currently not yet possible to configure specific sensors to operate at specific frequencies. This feature is coming soon but not available today.
|Specification||vCPU||RAM (GB)||SSD (TB)||EBS||EC2||S3||Data Transfer||Total Monthly||Total Yearly|
|"10 participants, active only"||2 (t3.small)||2||0.5||50||13.72||0||0.48||64.2||770.4|
|"100 participants, active only"||2 (t3.small)||2||0.5||50||13.72||0||2.06||65.78||789.36|
|"1000 participants, active only"||2 (t3.small)||2||1||102.4||13.72||0||2.06||118.18||1418.16|
|"10 participants, active + passive"||4 (c5.large)||8||1||102.4||124.1||23.56||5.12||255.18||3062.16|
|"100 participants, active + passive"||4 (c5d.large)||16||2||204.8||186.4||23.56||10.24||425||5100.000000000001|
|1000 participants active + passive||8 (c5d.2xlarge)||32||4||409.6||248.2||23.56||20.48||701.84||8422.079999999998|
When self-hosting the LAMP Platform (that is, "on-prem"), it's difficult to determine the monthly costs and maintenance needs. Please reach out to us for a consultation if you or your organization would like assistance with self-hosting.
Though self-hosting may appear to be cheaper, the "hidden costs" should not be neglected in: ensuring redundant storage, no system down-time, regular storage backups, HIPAA-compatible encryption during flight and at rest, preventing network bottlenecks, regular system maintenance, and more.
We do not recommend self-hosting unless your organization already has the infrastructure to correctly do so.