eduKateSG · Why Science?
Learn from the first careful clinical use while protecting every participant
Follow FDA’s October 2013 final guidance through proof of principle, initial safety, nonclinical gaps, risk controls, monitoring, device changes and transparent learning.
Some device questions cannot be answered fully on a bench, in a model or in an animal study. A carefully limited early feasibility study can provide the first clinical information needed to refine a design, understand device function and plan later investigations.
FDA’s October 2013 final guidance addresses IDE applications for early feasibility studies of significant-risk devices, including certain first-in-human studies. FDA says these studies can provide proof of principle and initial clinical safety data when clinical experience is necessary because nonclinical methods are unavailable or inadequate.
Did you know? “Early” describes a development stage, not a lighter duty to protect people. FDA states that initiation still requires an appropriate benefit-risk analysis and adequate human-subject protection measures.
This is science education at its most responsible: uncertainty is named, learning goals are narrow and every risk has a control. The same habits support research skills and future work in engineering, clinical medicine, ethics and regulation.
Chapter index
Chapters 1–9: early-feasibility purpose, participants and benefit-risk
- 1. An EFS is a limited early clinical investigation
- 2. The October 2013 guidance is final
- 3. Clinical need should justify moving beyond nonclinical work
- 4. Proof of principle and initial safety are distinct aims
- 5. Benefit-risk analysis begins with remaining uncertainty
- 6. Participant selection should fit the learning question
- 7. Informed consent should explain uncertainty plainly
- 8. The device description and version must be exact
- 9. Training and site readiness are risk controls
Chapters 10–17: controls, monitoring, device modification and development decisions
- 10. Monitoring should support rapid learning
- 11. Stopping rules protect participants
- 12. Device modifications need controlled iteration
- 13. Protocol changes should preserve interpretability
- 14. Adverse events and device deficiencies teach different lessons
- 15. Follow-up length should match plausible risks
- 16. Early FDA interaction can focus the IDE plan
- 17. A complete EFS receipt preserves every learning step
CHAPTER 1 OF 17 · Define early learning
1. An EFS is a limited early clinical investigation
FDA describes an early feasibility study as a limited clinical investigation of a device early in development. It typically involves a small number of participants and evaluates the design concept with respect to initial clinical safety and device functionality. Results may guide device modifications before a larger feasibility or pivotal programme.
For an efs is a limited early clinical investigation, learning and protection must move together. FDA describes an early feasibility study as a limited clinical investigation of a device early in development. It typically involves a small number of participants and evaluates the design concept with respect to initial clinical safety and device functionality. Results may guide device modifications before a larger feasibility or pivotal programme. Keep this boundary visible: Small size and early timing limit the conclusions that can be drawn.
Try this
Create a risk-to-control map for an efs is a limited early clinical investigation. State the device-development decision that requires human clinical information now. Add who monitors the control and what stops enrolment. Preserve this limit: Small size and early timing limit the conclusions that can be drawn.
A strong science note separates observation from interpretation. State the device-development decision that requires human clinical information now. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
State the device-development decision that requires human clinical information now. Keep this boundary visible: Small size and early timing limit the conclusions that can be drawn.
The final guidance is dated October 2013 and carries docket FDA-2011-D-0787. It focuses on IDE applications for early feasibility studies of significant-risk devices, including certain first-in-human studies. The document supports early clinical evaluation in the United States under the IDE framework.
Early evidence can be exciting precisely because it is new. In the october 2013 guidance is final, The final guidance is dated October 2013 and carries docket FDA-2011-D-0787. It focuses on IDE applications for early feasibility studies of significant-risk devices, including certain first-in-human studies. The document supports early clinical evaluation in the United States under the IDE framework. The honest stopping point is: The guidance is not a general waiver of IDE, IRB or informed-consent responsibilities.
Try this
Make a learning plan for the october 2013 guidance is final. Create a source receipt with scope, final status, date, docket and target study type. Separate information needed for the next participant from information useful only for later development. Keep this boundary: The guidance is not a general waiver of IDE, IRB or informed-consent responsibilities.
A strong science note separates observation from interpretation. Create a source receipt with scope, final status, date, docket and target study type. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a source receipt with scope, final status, date, docket and target study type. Keep this boundary visible: The guidance is not a general waiver of IDE, IRB or informed-consent responsibilities.
CHAPTER 3 OF 17 · Define early learning
3. Clinical need should justify moving beyond nonclinical work
An EFS may be appropriate when clinical experience is needed because available nonclinical methods cannot provide the information required to advance development. The package should explain what bench, animal, computational or other work has shown, what it cannot answer and why the remaining question matters.
Treat clinical need should justify moving beyond nonclinical work as a carefully controlled first look, with each uncertainty linked to a safeguard. An EFS may be appropriate when clinical experience is needed because available nonclinical methods cannot provide the information required to advance development. The package should explain what bench, animal, computational or other work has shown, what it cannot answer and why the remaining question matters. Remember: A desire for faster development is not enough to justify unresolved avoidable risk.
Try this
Use a participant-first review for clinical need should justify moving beyond nonclinical work. Build a gap table listing each question, completed nonclinical evidence and why clinical observation is necessary. Ask what is uncertain, how it is explained and which safeguard responds. Respect this limit: A desire for faster development is not enough to justify unresolved avoidable risk.
A strong science note separates observation from interpretation. Build a gap table listing each question, completed nonclinical evidence and why clinical observation is necessary. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Build a gap table listing each question, completed nonclinical evidence and why clinical observation is necessary. Keep this boundary visible: A desire for faster development is not enough to justify unresolved avoidable risk.
CHAPTER 4 OF 17 · Define early learning
4. Proof of principle and initial safety are distinct aims
Proof of principle asks whether the device can perform its intended clinical function; initial safety looks for early harms and the practical conditions of use. The protocol should distinguish these objectives and select observations that can inform each. An encouraging functional signal does not erase a safety concern.
For proof of principle and initial safety are distinct aims, learning and protection must move together. Proof of principle asks whether the device can perform its intended clinical function; initial safety looks for early harms and the practical conditions of use. The protocol should distinguish these objectives and select observations that can inform each. An encouraging functional signal does not erase a safety concern. Keep this boundary visible: EFS results usually do not establish definitive safety or effectiveness.
Try this
Create a risk-to-control map for proof of principle and initial safety are distinct aims. Write separate success and concern criteria for function and safety. Add who monitors the control and what stops enrolment. Preserve this limit: EFS results usually do not establish definitive safety or effectiveness.
A strong science note separates observation from interpretation. Write separate success and concern criteria for function and safety. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Write separate success and concern criteria for function and safety. Keep this boundary visible: EFS results usually do not establish definitive safety or effectiveness.
CHAPTER 5 OF 17 · Build protections
5. Benefit-risk analysis begins with remaining uncertainty
The benefit-risk case should consider the condition, available alternatives, anticipated benefit, known and potential risks, uncertainty and safeguards. Participants may have serious conditions or limited options, but vulnerability increases the need for careful explanation and oversight. The reasoning should connect each uncertainty to a control or monitoring action.
Early evidence can be exciting precisely because it is new. In benefit-risk analysis begins with remaining uncertainty, The benefit-risk case should consider the condition, available alternatives, anticipated benefit, known and potential risks, uncertainty and safeguards. Participants may have serious conditions or limited options, but vulnerability increases the need for careful explanation and oversight. The reasoning should connect each uncertainty to a control or monitoring action. The honest stopping point is: High unmet need does not make every device risk acceptable.
Try this
Make a learning plan for benefit-risk analysis begins with remaining uncertainty. Create a risk register with severity, likelihood, detectability, mitigation and residual risk. Separate information needed for the next participant from information useful only for later development. Keep this boundary: High unmet need does not make every device risk acceptable.
A strong science note separates observation from interpretation. Create a risk register with severity, likelihood, detectability, mitigation and residual risk. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a risk register with severity, likelihood, detectability, mitigation and residual risk. Keep this boundary visible: High unmet need does not make every device risk acceptable.
CHAPTER 6 OF 17 · Build protections
6. Participant selection should fit the learning question
Eligibility should identify people for whom the potential benefit-risk balance is supportable and whose clinical situation can answer the study objective. Narrow criteria may reduce uncertainty in first use, while exclusions can protect people at particular risk. Recruitment should avoid overstating benefit in a study designed mainly for learning.
Treat participant selection should fit the learning question as a carefully controlled first look, with each uncertainty linked to a safeguard. Eligibility should identify people for whom the potential benefit-risk balance is supportable and whose clinical situation can answer the study objective. Narrow criteria may reduce uncertainty in first use, while exclusions can protect people at particular risk. Recruitment should avoid overstating benefit in a study designed mainly for learning. Remember: The enrolled group may not represent later intended users.
Try this
Use a participant-first review for participant selection should fit the learning question. Map each inclusion and exclusion criterion to either scientific interpretability or participant protection. Ask what is uncertain, how it is explained and which safeguard responds. Respect this limit: The enrolled group may not represent later intended users.
A strong science note separates observation from interpretation. Map each inclusion and exclusion criterion to either scientific interpretability or participant protection. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Map each inclusion and exclusion criterion to either scientific interpretability or participant protection. Keep this boundary visible: The enrolled group may not represent later intended users.
CHAPTER 7 OF 17 · Build protections
7. Informed consent should explain uncertainty plainly
Consent should communicate the experimental nature of the device, limited human experience, alternatives, foreseeable risks, procedures, follow-up and possibility of device changes or study stopping. Comprehension matters more than document length. Discussion should leave room for questions and voluntary choice.
For informed consent should explain uncertainty plainly, learning and protection must move together. Consent should communicate the experimental nature of the device, limited human experience, alternatives, foreseeable risks, procedures, follow-up and possibility of device changes or study stopping. Comprehension matters more than document length. Discussion should leave room for questions and voluntary choice. Keep this boundary visible: A signed form alone does not demonstrate informed understanding.
Try this
Create a risk-to-control map for informed consent should explain uncertainty plainly. Rewrite the largest unknown in plain language and design a teach-back question. Add who monitors the control and what stops enrolment. Preserve this limit: A signed form alone does not demonstrate informed understanding.
A strong science note separates observation from interpretation. Rewrite the largest unknown in plain language and design a teach-back question. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Rewrite the largest unknown in plain language and design a teach-back question. Keep this boundary visible: A signed form alone does not demonstrate informed understanding.
CHAPTER 8 OF 17 · Build protections
8. The device description and version must be exact
The IDE package should identify the device, components, materials, software, manufacturing state and relevant testing. Early devices may evolve rapidly, so version control is essential. Investigators need to know which configuration each participant received and which evidence supported that configuration.
Early evidence can be exciting precisely because it is new. In the device description and version must be exact, The IDE package should identify the device, components, materials, software, manufacturing state and relevant testing. Early devices may evolve rapidly, so version control is essential. Investigators need to know which configuration each participant received and which evidence supported that configuration. The honest stopping point is: Evidence for one prototype cannot be assumed to cover a materially changed version.
Try this
Make a learning plan for the device description and version must be exact. Create a configuration record linking serial number, software, procedure, participant and evidence package. Separate information needed for the next participant from information useful only for later development. Keep this boundary: Evidence for one prototype cannot be assumed to cover a materially changed version.
A strong science note separates observation from interpretation. Create a configuration record linking serial number, software, procedure, participant and evidence package. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Create a configuration record linking serial number, software, procedure, participant and evidence package. Keep this boundary visible: Evidence for one prototype cannot be assumed to cover a materially changed version.
Early use may depend heavily on investigator skill, imaging, procedural planning, rescue capability or specialised follow-up. Training, proctoring, facility requirements and emergency plans should match the failure modes. Readiness should be documented before the first procedure, not inferred from professional titles.
Treat training and site readiness are risk controls as a carefully controlled first look, with each uncertainty linked to a safeguard. Early use may depend heavily on investigator skill, imaging, procedural planning, rescue capability or specialised follow-up. Training, proctoring, facility requirements and emergency plans should match the failure modes. Readiness should be documented before the first procedure, not inferred from professional titles. Remember: Experienced clinicians still need device-specific preparation.
Try this
Use a participant-first review for training and site readiness are risk controls. Run a simulated procedure and record deviations, rescue steps and retraining needs. Ask what is uncertain, how it is explained and which safeguard responds. Respect this limit: Experienced clinicians still need device-specific preparation.
A strong science note separates observation from interpretation. Run a simulated procedure and record deviations, rescue steps and retraining needs. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Run a simulated procedure and record deviations, rescue steps and retraining needs. Keep this boundary visible: Experienced clinicians still need device-specific preparation.
Close monitoring can include real-time procedural observations, frequent follow-up, device performance checks and prompt adverse-event review. The plan should define what data are reviewed, by whom and how quickly. Escalation criteria should trigger pausing, additional testing or protocol changes when new risk appears.
For monitoring should support rapid learning, learning and protection must move together. Close monitoring can include real-time procedural observations, frequent follow-up, device performance checks and prompt adverse-event review. The plan should define what data are reviewed, by whom and how quickly. Escalation criteria should trigger pausing, additional testing or protocol changes when new risk appears. Keep this boundary visible: Frequent monitoring is useful only when thresholds and authority are clear.
Try this
Create a risk-to-control map for monitoring should support rapid learning. Write one safety trigger with data source, threshold, decision-maker and maximum response time. Add who monitors the control and what stops enrolment. Preserve this limit: Frequent monitoring is useful only when thresholds and authority are clear.
A strong science note separates observation from interpretation. Write one safety trigger with data source, threshold, decision-maker and maximum response time. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Write one safety trigger with data source, threshold, decision-maker and maximum response time. Keep this boundary visible: Frequent monitoring is useful only when thresholds and authority are clear.
Prespecified stopping or pausing criteria can respond to serious adverse events, unexpected device failures, repeated technical problems or inability to follow the protocol. The rule should distinguish an automatic pause from a review trigger and describe restart requirements. The small sample makes each event especially informative but also statistically unstable.
Early evidence can be exciting precisely because it is new. In stopping rules protect participants, Prespecified stopping or pausing criteria can respond to serious adverse events, unexpected device failures, repeated technical problems or inability to follow the protocol. The rule should distinguish an automatic pause from a review trigger and describe restart requirements. The small sample makes each event especially informative but also statistically unstable. The honest stopping point is: Absence of an event in a few participants does not prove low risk.
Try this
Make a learning plan for stopping rules protect participants. Test the stopping rule against three hypothetical events and document the authorised action. Separate information needed for the next participant from information useful only for later development. Keep this boundary: Absence of an event in a few participants does not prove low risk.
A strong science note separates observation from interpretation. Test the stopping rule against three hypothetical events and document the authorised action. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Test the stopping rule against three hypothetical events and document the authorised action. Keep this boundary visible: Absence of an event in a few participants does not prove low risk.
An EFS may reveal design changes needed for safety or function. A change record should describe the problem, proposed modification, engineering assessment, verification, new risk analysis and whether FDA or IRB review is needed before further use. Iteration is a purpose of EFS, but it must not become undocumented experimentation.
Treat device modifications need controlled iteration as a carefully controlled first look, with each uncertainty linked to a safeguard. An EFS may reveal design changes needed for safety or function. A change record should describe the problem, proposed modification, engineering assessment, verification, new risk analysis and whether FDA or IRB review is needed before further use. Iteration is a purpose of EFS, but it must not become undocumented experimentation. Remember: Learning from one version does not automatically validate the next version.
Try this
Use a participant-first review for device modifications need controlled iteration. Trace a proposed change from clinical observation through risk assessment, testing and authorisation. Ask what is uncertain, how it is explained and which safeguard responds. Respect this limit: Learning from one version does not automatically validate the next version.
A strong science note separates observation from interpretation. Trace a proposed change from clinical observation through risk assessment, testing and authorisation. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Trace a proposed change from clinical observation through risk assessment, testing and authorisation. Keep this boundary visible: Learning from one version does not automatically validate the next version.
CHAPTER 13 OF 17 · Carry evidence forward
13. Protocol changes should preserve interpretability
Changes to technique, eligibility, follow-up or endpoints may be necessary as experience grows. Amendments should be dated, justified and linked to evidence. Analyses should identify which participants were treated under which version so differences are not averaged into a misleading single result.
For protocol changes should preserve interpretability, learning and protection must move together. Changes to technique, eligibility, follow-up or endpoints may be necessary as experience grows. Amendments should be dated, justified and linked to evidence. Analyses should identify which participants were treated under which version so differences are not averaged into a misleading single result. Keep this boundary visible: Flexible learning does not justify rewriting outcomes after seeing favourable cases.
Try this
Create a risk-to-control map for protocol changes should preserve interpretability. Build a participant-by-version matrix for device, procedure and protocol amendments. Add who monitors the control and what stops enrolment. Preserve this limit: Flexible learning does not justify rewriting outcomes after seeing favourable cases.
A strong science note separates observation from interpretation. Build a participant-by-version matrix for device, procedure and protocol amendments. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Build a participant-by-version matrix for device, procedure and protocol amendments. Keep this boundary visible: Flexible learning does not justify rewriting outcomes after seeing favourable cases.
CHAPTER 14 OF 17 · Carry evidence forward
14. Adverse events and device deficiencies teach different lessons
Clinical harms, near misses, malfunctions and device deficiencies can reveal distinct failure pathways. Reports should preserve timing, device version, procedure, contributing factors, treatment and outcome. Root-cause analysis should avoid attributing every problem to user error without examining design and training.
Early evidence can be exciting precisely because it is new. In adverse events and device deficiencies teach different lessons, Clinical harms, near misses, malfunctions and device deficiencies can reveal distinct failure pathways. Reports should preserve timing, device version, procedure, contributing factors, treatment and outcome. Root-cause analysis should avoid attributing every problem to user error without examining design and training. The honest stopping point is: A non-injurious deficiency may still predict serious future risk.
Try this
Make a learning plan for adverse events and device deficiencies teach different lessons. Write separate clinical-event and device-performance narratives for one hypothetical failure. Separate information needed for the next participant from information useful only for later development. Keep this boundary: A non-injurious deficiency may still predict serious future risk.
A strong science note separates observation from interpretation. Write separate clinical-event and device-performance narratives for one hypothetical failure. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Write separate clinical-event and device-performance narratives for one hypothetical failure. Keep this boundary visible: A non-injurious deficiency may still predict serious future risk.
CHAPTER 15 OF 17 · Carry evidence forward
15. Follow-up length should match plausible risks
Immediate procedural success may not reveal healing, migration, durability, infection or delayed functional problems. Follow-up schedules should reflect device mechanism, materials, implantation and anticipated failure modes. Lost follow-up is especially consequential in a small study and should be actively addressed.
Treat follow-up length should match plausible risks as a carefully controlled first look, with each uncertainty linked to a safeguard. Immediate procedural success may not reveal healing, migration, durability, infection or delayed functional problems. Follow-up schedules should reflect device mechanism, materials, implantation and anticipated failure modes. Lost follow-up is especially consequential in a small study and should be actively addressed. Remember: Short observation cannot support claims about long-term performance.
Try this
Use a participant-first review for follow-up length should match plausible risks. Map each risk to the earliest and latest time it could appear and set observation windows. Ask what is uncertain, how it is explained and which safeguard responds. Respect this limit: Short observation cannot support claims about long-term performance.
A strong science note separates observation from interpretation. Map each risk to the earliest and latest time it could appear and set observation windows. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Map each risk to the earliest and latest time it could appear and set observation windows. Keep this boundary visible: Short observation cannot support claims about long-term performance.
CHAPTER 16 OF 17 · Carry evidence forward
16. Early FDA interaction can focus the IDE plan
Sponsors can use early discussion to address nonclinical evidence, acceptable risk controls, study objectives, monitoring, device changes and reporting. Focused questions and a clear evidence map make feedback more useful. The agency’s willingness to discuss an approach does not replace sponsor responsibility or final review.
For early fda interaction can focus the ide plan, learning and protection must move together. Sponsors can use early discussion to address nonclinical evidence, acceptable risk controls, study objectives, monitoring, device changes and reporting. Focused questions and a clear evidence map make feedback more useful. The agency’s willingness to discuss an approach does not replace sponsor responsibility or final review. Keep this boundary visible: Pre-submission feedback is not IDE approval.
Try this
Create a risk-to-control map for early fda interaction can focus the ide plan. Prepare three decision-changing questions and the supporting evidence for each. Add who monitors the control and what stops enrolment. Preserve this limit: Pre-submission feedback is not IDE approval.
A strong science note separates observation from interpretation. Prepare three decision-changing questions and the supporting evidence for each. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Prepare three decision-changing questions and the supporting evidence for each. Keep this boundary visible: Pre-submission feedback is not IDE approval.
CHAPTER 17 OF 17 · Carry evidence forward
17. A complete EFS receipt preserves every learning step
The receipt records device and protocol versions, nonclinical evidence, residual risks, participant criteria, consent, training, monitoring, stopping decisions, adverse events, deficiencies, modifications, approvals and next-development decisions. It lets later teams understand what was learned without exaggerating an early signal.
Early evidence can be exciting precisely because it is new. In a complete efs receipt preserves every learning step, The receipt records device and protocol versions, nonclinical evidence, residual risks, participant criteria, consent, training, monitoring, stopping decisions, adverse events, deficiencies, modifications, approvals and next-development decisions. It lets later teams understand what was learned without exaggerating an early signal. The honest stopping point is: An EFS receipt documents initial learning; it cannot support conclusions beyond the study.
Try this
Make a learning plan for a complete efs receipt preserves every learning step. Ask an independent reviewer to identify which questions are answered, unresolved and newly created. Separate information needed for the next participant from information useful only for later development. Keep this boundary: An EFS receipt documents initial learning; it cannot support conclusions beyond the study.
A strong science note separates observation from interpretation. Ask an independent reviewer to identify which questions are answered, unresolved and newly created. Record the source, version or date, population or device context, method, result and one uncertainty. If the official record does not show a field, write “not shown” instead of filling the gap from memory.
Evidence checkpoint
Ask an independent reviewer to identify which questions are answered, unresolved and newly created. Keep this boundary visible: An EFS receipt documents initial learning; it cannot support conclusions beyond the study.
Early-feasibility evidence and protection map
| Question | Evidence to capture | Boundary to retain |
|---|---|---|
| Why human study now? | Nonclinical evidence, remaining gap and development decision | Speed alone is not justification |
| Who may participate? | Eligibility, alternatives, consent and anticipated benefit-risk | Small samples limit generalisation |
| How are risks controlled? | Training, monitoring, stopping and rescue plans | No-event results do not prove safety |
| How is learning carried forward? | Versions, modifications, events and next steps | Initial evidence is not definitive effectiveness |
Science learning becomes useful when a student can show where a claim came from, why it applies and where it stops. Continue with the eduKate Science Learning Hub, compare choices in the Education Hub, and connect these habits to career pathways built from subject capabilities.
An early feasibility study is hopeful because careful first use can turn uncertainty into better design. Its value comes from pairing every learning objective with protection, every modification with evidence and every promising result with a visible limit.
Primary sources and verification date
Source status checked 25 October 2026. FDA guidance describes the agency’s current thinking and does not replace applicable statutes, regulations or product-specific communication with FDA.
- FDA: IDEs for Early Feasibility Medical Device Clinical Studies — Final guidance, October 2013.
- FDA: Early Feasibility Studies Program — Current programme resources.
- FDA: IDE approval process — Current EFS description and IDE pathway context.