This research study, Personalized, Responsive Intervention Sequences for Minimally Verbal Children with Autism (PRISM), is designed to maximize language outcomes for limited-language preschoolers, thereby lowering the risk of being classified as "minimally verbal" at age 6, by empirically developing a two-stage, 20-week adaptive intervention approach in a real world community settings. If found efficacious, the adaptive intervention design will capitalize on the heterogeneity and evolving status of children with ASD by providing the best intervention (DTT, JASPER and CET) for children who need it (leading to individualized sequences of intervention), only when it is needed (potentially reducing burden on children).

This study will use a Sequential Multiple Assignment Randomized Trial (SMART) design to examine key scientific questions for assembling an optimal adaptive intervention for preschoolers with ASD .

Children will first be recruited through local school districts and early intervention service providers, approximately 35 children per year across all three sites (UCLA will have 15 per year, University of Oregon will have 10 per year, and University of Rochester will have 10 per year) for a total of 140 children.

Based on early intervention research conducted by this research team over the past 15 years, we are proposing a first stage treatment decision comparing the commonly applied ABA method of Discrete Trial Teaching or an NDBI approach of JASPER focused on social communication and play targets specifically. The dose of each intervention is delivered at a research dose of 2 hours per week (consistent with previous JASPER studies. All children are expected to receive federally subsidized early intervention services which we will track for dose and type (to be considered in analyses). The interventions are described below: Stage 1 DTT Intervention (2x60 minute sessions per week in community setting): Discrete Trial Training (DTT), an adult-led, highly structured, behavioral teaching approach, is considered to have the strongest evidence as a "standard of care" for young children with autism. DTT emphasizes didactic, adult-led instruction. This approach relies on teaching discriminations between stimuli, responses to stimuli, and providing systematic reinforcement for correct responses. The goals of DTT are to teach specific skills, to accelerate overall development and increase school readiness. In order to meet these goals the interventionist works to create an engaging environment to support the child's ability to attend to the instruction and demonstrate the skills correctly. Recently, DTT interventions have begun to teach skills to address early core social communication deficits such as joint attention in order to promote better language and communication outcomes. While many children will have been exposed to DTT prior to entering this trial as part of business as usual community treatment, community intervention quality and dosage can vary widely. It is important to insure that children (a) receive quality DTT, and (b) have exposure to elements related to language learning, specifically engagement in instruction, and joint attention and requesting gestures, in order to make the comparison with JASPER in stage 1. Thus, in addition to the expected ABA approach used in their early intervention preschool program (which in the sites we are located also includes DTT during the day), we will insure high quality DTT carried out by research staff in order to adequately make the comparison between models for children with ASD in this project.

Stage 1 JASPER Intervention (2x60 minute sessions per week in community setting): is a developmentally anchored behavioral intervention that assumes that communication develops from social interactions in which specific social engagement strategies, symbolic representations, and early communication forms are modeled and naturally reinforced by the adult's responses to the child. JASPER uses the child's current play level to choose appropriate toys and materials to create a context for learning. Developmental principles of following the child's attentional focus, balancing imitation and modeling to create and expand play routines, supporting children's regulation and engagement, as well as responding to and expanding children's communication/language targets are used. The goal of JASPER is to increase (a) joint engagement, (b) initiations of joint attention and requesting skills, (c) diversity in social play involving objects and persons, and (d) verbal and nonverbal communication by facilitating meaningful social interactions. The social interaction foundation of JASPER is critical. Modeling and expansions of communicative behaviors and play are used strategically within the interactions with an adult. For children at risk of minimally verbal status, meaningful social interaction is essential for establishing the platform on which language input and development will be built. Unlike DTT, JASPER is likely to be a novel intervention to which few children will have had previous exposure through business as usual treatment.

Stage 2: Responders (2x60 minute sessions per week): Responders in stage 1 stay the course in the same intervention at the same dose.

Stage 2: Intensify (4x60 minute sessions per week): Slower responders in stage 1 may be rerandomized to continue the stage 1 intervention. In this Stage 2 choice, we will follow the same procedures outlined in the stage 1 interventions to provide the child with more time to make gains with consistent high quality intervention, but add an additional session, as increased dose is usual practice in the community.

Stage 2 intervention Combined & Enhanced Treatment (CET: 2x60 minute sessions per week): CET is 2 sessions per week. Using the dashboard approach pioneered by Chorpita, child needs are matched to potentially effective intervention modules (e.g., behavior challenges that may indicate more time spent in DTT type intervention). We have used this approach with older minimally verbal children with ASD, and are proposing to use this model here with preschool aged children who respond slowly to first stage intervention.

Our prior trials exploring supports for slowly responding school age children have shown us that children are typically showing some response rather than no response. Therefore, CET will combine a balance of initial treatment augmented with additional treatment type. Our options for enhancing intervention include a) primarily direct instruction (75% DTT with JASPER strategies embedded within play based breaks), b)primarily JASPER (75% JASPER, with DTT priming on specific language or play targets at the beginning of the session and trials embedded briefly within JASPER routines), c) add SGD, child will receive access to a tablet with a speech generating application to teach the child to communicate using both augmented and spoken language using either JASPER or direct instruction strategies. When the SGD is added, this will provide another mode for children to communicate as well to model and teach expressive communication. The SGD will be added to DTT to provide instruction (e.g., discriminative stimulus) and for the child to use to respond and spontaneously communicate across programs. In JASPER, the adult will model both augmented words using the SGD and spoken words as well as respond and expand child communication with both spoken and augmented words.

The tablet will be programmed to meet the developmental level of the child (e.g., number of symbols, type of symbol- real image, symbol). Children randomized to CET will be reviewed by the study team (site leads, coordinators, and interventionist) to plan the CET program components to cross site consensus. Plans and components will be tracked using the CET Dashboard tracking form.

Intervention Dose and Duration: In stage 1, all children receive 2 hours per week of JASPER or DTT.

The total duration of treatment is 20 weeks This dose is based on (i) ongoing and previous research conducted by our team concerning feasible doses of stage 1 DTT or JASPER in actual practice settings (RO1 # R01HD073975); and (ii) our previous efficacy studies of JASPER. Concomitant Intervention: Children can receive concomitant intervention, which will be measured for dose and type (and considered in analyses). For example, many (or all) children are expected to receive federally subsidized early intervention services.

Fidelity and Quality Assurance. Interventionists at all 3 recruitment sites will be trained together to high fidelity levels (>90% on all elements of intervention components) with 2-3 children prior to carrying out the intervention for this study. The PI and Coordinator at each site will supervise interventionists weekly in individual and group meetings. Coordinators will rate fidelity for each site for 20% of all sessions, selected randomly. If low levels of fidelity are observed, the poor-performing interventionist will receive additional training until fidelity exceeds 90%. Weekly conference calls will be conducted to monitor cross-site fidelity for each intervention.

Video-recorded sessions of each interventionist will be reviewed on a rotating basis by the interventionists and supervisors during these calls. In prior studies we have successfully used this approach in which videos are uploaded to our password-protected, HIPAA-compliant Data Coordinating site to calibrate reliability ratings and intervention fidelity and to initiate discussion of difficult or problematic cases in prior studies. Quality assurance checks at each site will occur monthly on 10% of randomly selected sessions by Coordinators and/or PIs.

Randomization. Authorized personnel at each site will randomize participants via the centralized database using a customized system tailored for the SMART design. The project statistician will generate the randomization sequence, with allocation sequence concealed from all other study personnel. The first randomization at baseline is to stage 1 treatment (JASPER or DTT) with equal assignment probability to each group. For the first randomization, the system will make a treatment assignment only if the participant meets study criteria and only after the screening and stratification data are entered. For the second randomization, once the variables necessary for determining early response status (as well as the stratification variables) are entered, slower responders will be automatically re-randomized to Stage 2 treatment.

All randomizations will occur within sites and will be further stratified using a minimization allocation method. This procedure will ensure that treatment groups are balanced for variables that may correlate highly with outcomes. The first randomization will be stratified on 1 baseline measure, initiating joint attention (JA gestures and/or JA language >0 vs ≤0) based on the Early Social Communication Scales (ESCS). The second randomization among slow responders will be stratified on whether or not the child has made 0.2 gains (2 minutes) in joint engagement from baseline to week 10 based on the Adult-Child Interaction Play Interaction.

SMART Design:

Early Assessments, Randomize to Stage 1 Treatment

Stage 1: JASPER

Decision Point: Early Response Assessments

Early Responders/Re-Randomize JASPER (slow responders)

Stage 2: Stay the course (Early Responders only)- (A; look at 6.0) Stage 2:CET - (B; look at 6.0) or Stage 2: JASPER -intensify (Randomized for slow responders)- (C; look at 6.0)

Exit assessments and Follow up assessments at 10 weeks post treatment and 6 years of age

Early Assessments, Randomize to Stage 1 Treatment

Stage 1: DTT Decision Point: Early Response Assessments

Early Responders/Re-Randomize DTT (slow responders)

Stage 2: Stay the course (Early Responders only) -(D; look at 6.0) Stage 2:CET -(E; look at 6.0) or Stage 2: DTT -intensify (Randomized for slow responders) - (F; look at 6.0)

Exit assessments and Follow up assessments at 10 weeks post treatment and 6 years of age

Inclusion Criteria:

• children meeting ADOS-2 criteria for ASD,
• age 36-59 months
• who have had > 3 months early intervention/preschool (to ensure that children already have been exposed to some community interventions) and
• use < 20 functional words (i.e., non-echoed, non-scripted).

Additional inclusion criteria are:

• stable medication over the past 6 months, and
• nonverbal mental age of >12 months on the Mullen Scales of Early Learning (visual reception and fine motor subscales).

Exclusion Criteria:

• We will exclude children who have sensory and motor impairments (e.g., visual impairment, deaf or hard of hearing) and with known genetic syndromes (e.g., Down Syndrome).
• We will not exclude on the basis of AAC exposure, but expect few will be exposed at these ages.