> For the complete documentation index, see [llms.txt](https://docs.nested-knowledge.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.nested-knowledge.com/review-workflow/screening/exclusion-based-screening/screening-model.md). # Screening Model Learn how to use the screening model to generate inclusion probabilities for each record. **Screening Model** assigns each record an inclusion probability from your team’s adjudicated Exclusion-Based Screening decisions. Use it to prioritize records in Single Screening or to power **Robot Screener** in Dual Screening. The model is available only with Exclusion-Based Screening. In Criteria-Based Screening, **Smart Screener** answers defined eligibility questions instead. See [Criteria-Based Screening](/review-workflow/screening/criteria-based-screening.md) to choose the paradigm that fits your protocol. An inclusion probability is a prioritization signal, not a final decision. It can help you decide which records to review first. ### Set Up Screening Model {% hint style="info" %} Confirm your nest uses **Exclusion-Based Screening**. **Screening Model** does not work with **Criteria-Based Screening**. {% endhint %} {% stepper %} {% step %} #### Open Nest Settings On **Nest Settings**, under **Screening Model**, turn on **Automatic Training**. With **Automatic Training** on, the model retrains after every 10 newly screened records.

Automatic Training setting under Screening Model in Nest Settings. — Turn on Automatic Training in Nest Settings.

{% endstep %} {% step %} #### Meet The Threshold The model begins assigning probabilities after 50 screened records, including 10 included or advanced records. {% endstep %} {% step %} #### View Inclusion Probabilities {% hint style="info" %} **Advancement probability** estimates whether a record advances from **Abstract Screening** to **Full Text Screening**. It applies only in **Two Pass Screening**. **Inclusion probability** estimates whether a record will be included. It applies in every screening mode. {% endhint %} On **Screening** and **Study Inspector**, probabilities appear above the exclusion reason list. Before the model trains, the record shows **Train Screening Model**.

Inclusion probability above the exclusion reason list. — Inclusion probability appears above the exclusion reason list.

{% endstep %} {% step %} #### Sort The Queue By Inclusion Probability On **Nest Settings**, under **Blinding → Queue Ordering**, select **Likelihood of Inclusion**. The queue shows records with the highest inclusion probability first. You can hide probabilities in **Screening** to reduce bias.

Likelihood of Inclusion queue ordering in Nest Settings. — Select Likelihood of Inclusion to prioritize records with higher probabilities.

{% endstep %} {% step %} #### Filter By Inclusion Probability In [Study Inspector](/review-workflow/study-inspector.md), select **Add Filter**, then select **Advancement/Inclusion Probability**. Select **Recommend: Advance/Include**, **Recommend: Exclude**, or enter a custom range.

Advancement and inclusion probability filter in Study Inspector. — Filter records by advancement or inclusion probability in Study Inspector.

{% endstep %} {% endstepper %} ### Interpret The Probability Scores A higher score means the record resembles records your team previously included or advanced. A lower score means it resembles records your team excluded. Treat every score as a prioritization signal. A low-probability record can still belong in your review. Every record still needs a human decision before finalization. Scores can change as the model retrains. ### Review Model Performance To review predictions and cross-validation metrics, select an inclusion probability on any record. You can also open **Nest Settings**, then select **Screening → View Screening Model**.

Predictions

The **Predictions** tab charts Included or Advanced, Excluded, and Unscreened records. The x-axis shows inclusion or advancement probability from 0 to 1. Performance varies by nest because the model trains on your team’s decisions.

Cross Validations

The **Cross Validation** tab shows evaluation metrics, including recall and accuracy. Use these metrics with your protocol and human review process when deciding whether to use the model.

### Understand Model Limits Model performance depends on the quantity and quality of prior decisions in your nest. Early in a review, limited training data, incomplete metadata, or abstract-only information can reduce reliability. Review the model’s metrics before relying on probabilities. The [Screening Model model card](/ai-transparency/model-cards/screening.md#screening-model-robot-screener) documents its evaluation, input data, and limitations. ### FAQs

Does the screening model require Dual Screening?

No. In **Single Screening**, use it to sort or filter your queue. **Dual Screening** is required only for **Robot Screener**.

Is the inclusion probability the same thing as Robot Screener?

They use the same underlying model. In **Single Screening**, it produces a score that helps you prioritize work. In **Dual Screening**, **Robot Screener** fills one reviewer seat. A second human reviewer and a human adjudicator remain required.

Can I see why the model assigned a given probability?

The model provides a numerical score. It does not provide a record-level explanation.

Does switching to Criteria-Based Screening keep the model's training?

When you switch to **Criteria-Based Screening**, existing **Exclusion Reasons** decisions remain available if you switch back. The model does not train on new **Criteria-Based Screening** decisions.

### Next Steps


Screen Records	Review records in inclusion-probability order.	Screen Records
Robot Screener	Fill one Dual Screening reviewer seat.	Robot Screener

--- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.nested-knowledge.com/review-workflow/screening/exclusion-based-screening/screening-model.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.