Il tempo della Human Data Science Big Data & AI per l'ottimizazione della governace sanitaria - Luca Pinto - Principal RWI Motore Sanità Milano, 5 ...
←
→
Trascrizione del contenuto della pagina
Se il tuo browser non visualizza correttamente la pagina, ti preghiamo di leggere il contenuto della pagina quaggiù
Il tempo della Human Data Science
Big Data & AI per l’ottimizazione
della governace sanitaria
Luca Pinto – Principal RWI
Motore Sanità
Milano, 5 Dicembre 2019
Copyright © 2019 IQVIA. All rights reserved.
Per molti anni l’efficienza ha dettato l’agenda del SSN
Variazione spesa per Regioni in PdR vs non in PdR
Fondo Sanitario Evoluzione 4,0
Agenda
Nazionale del deficit 3,0
Perseguire
nel 2017
… i Piani di
€ Mld
2,0 Piani di Rientro
112,6 Rientro hanno
aiutato la ridu-
1,0
0,0
Lavorare Iso
Mld Euro Risorse
zione del deficit -1,0
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
complessivo Riqualificare la
CAGR1 = + 0,8% Regioni in PdR Regioni non in PdR
Rete
Ospedaliera
Costi del personale Riqualificazione
Attivare un
Ospedaliera
nel 2017
piano cronicità
integrato
34,3 1.159
Ospedali
Mld Euro pubblici
e privati
CAGR1 = - 0,9%
Fonte: Rapporto OASI 2016, Il Sole24Ore Sanità, Tavole SDO 2016, L'uso dei farmaci in Italia - Rapporto nazionale 2016, Monitoraggio della spesa sanitaria – Rapporto nazionale anno 2016
(1) Dal 2012
(2) Il divario tra il livello di finanziamento di «Patto per la Salute 2014-2016» e dopo il decreto di cui all'art. 1, co. 394, L'232/2016 è pari a 8,95 MLD di Euro nel 2018 e sarà pari a 10,94 MLD nel 2019 1
This presentation is confidential and must not be used for purposes other than those for which has been drawn up and must not be disclosed, reproduced and/or revealed to third parties without the prior written consent of IQVIA
Il futuro richiede un cambio di paradigma volto al Valore
Domanda Costi Offerta
Invecchiamento Innovazione Big Data
Reti d’offerta
Fragilità Aderenza Oncologica &
Ematologica
Lungo Multifunzionale
Sopravvivenza Prevenzione &
Multidisciplinare
© 2019 | IQVIA, All rights Reserved 2
La dinamica demografica assorbirà risorse crescenti
anche associata a crescente cronicità e co-morbidità
La mortalità è diminuita in misura significativa anche nel caso di diagnosi di tumore
Supravvivenza e cronicità, 2015-2020 [Mln Pazienti]
3,4 3,5
3,3 3,4
3,2
3,0
2,6
Patients [Mln]
2,4 2,5
2,2 2,3 2,3
2015 2016 2017 2018 2019 2020
Cancer survivors Total chronics
CANCER SURVIVORS TOTAL CHRONIC PATIENTS
Source: AIRTUM - AIFA, Analysis IQVIA
This presentation is confidential and must not be used for purposes other than those for which has been drawn up and must not be disclosed, reproduced and/or revealed to third parties without the prior written consent of IQVIA
Source: IQVIA
© IQVIA 2018. All rights reserved. 3
L’innovazione clinica accelera, a costi crescenti
La pipeline in fase avanzata è cresciuta del 60% negli utlimi 10 anni
Numero Molecole in Sviluppo Clinico ( 2007-2017) Nuovi Agenti in Sviluppo per Indicazione (2013-2017)
Il 2018 è stato record assoluto per numero 2 1
3
di nuove approvazioni di farmaci (59) 4 Head & Breast
Renal 5
Neck
Bladder Breast
1 6
Neuro-
blastoma Melanoma
2 3
Liver Colorectal
1 Lym-
Thyroid 9
phoma
Castel-
9 Leukimia
man’s 1
Lung Ovarian
11 4
Merkel
GIST
Cell
1 1
Sarcoma Cervical
3 Multiple Basal Cell 1
Myeloma Polycy- Carcinoma
Prostate themia
5 Gastric
Vera 1
Source: IQVIA Oncology Global Trends 2018; IQVIA Institute April 2018
1 2 1
© 2019 | IQVIA, All rights Reserved 4
Gestione di cronicità e aderenza liberano risorse importanti
La gestione del paziente con patologie croniche
Prevalenza Trattati* Aderenza
Ipertensione 27,6% 67,5% 55,1%
Solo il 41% dei 11-12 miliardi,
Ipercolesterolemia 7,2% 36,5% 43,1%
Ischemia 7,6% 50,0% 77,0% cittadini affetti da il possibile
Diabete 5,5% 100,0% 62,1% una patologia è risparmio
Asma-BPCO 5,9% 30,5% 14,2% regolarmente in derivante da una
Depressione 7,5% 31,0% 38,4%
cura maggiore presa in
Osteoporosi 5,2% 39,0% 46,3%
*In trattamento regolare Il grado di carico e maggiore
Fonte: OSMED; *In trattamento regolare
aderenza aderenza
terapeutica è terapeutica, in
media è del 45% termini di minori
costi sanitari di
Il 60% dei ricadute
cittadini non si
sente seguito nella
gestione della
patologia
Source: IQVIA
Source: IQVIA, 2016, involving more than 700 people >65 years old © IQVIA 2018. All rights reserved.
5
In questo scenario, Big Data e l’AI accellerano l’innovazione di
processo
Can accelerate the
4.8yrs to diagnosis Artificial
Intelligence
from onset of symptoms to first diagnosis Any technique that enables computers to
of a rare disease mimic human intelligence
NLP
A subset of
Can help find the Machine AI that
Learning enables
24% under-treated A subset of AI that enables computers
computers to improve at tasks to
of diabetics who are undiagnosed with experience understand
language
and untreated
Un/Supervis Neural
ed Learning Networks
Can help address the
60% non-adherent Deep Learning
patients on chronic therapy who are not refilling A subset of machine learning
that permits software to train
their prescription after 6 months itself to perform tasks by
exposing it to vast amounts
of data
6
© 2019 | IQVIA, All rights Reserved
...sul fronte clinico/scientifiche così come su quello dell’efficienza
e dell’ «engagement», dimensioni sempre più interconnesse
7
Il potenziale è enorme...Le sfide da affrontare e vincere sono quelle
della qualità dei dati, dell’interconnessione tra sistemi e piattaforme
e della corretta interpretazione clinico-sanitaria e di costo/efficacia
Unparallel Volume and Complexity, beyond any other industry
8
© 2019 | IQVIA, All rights Reserved
... senza menzionare il tema della verifica dei dati e della gestione
della data privacy che a loro volta hanno peculiarità uniche nel LS
Tech & Pharma giants invest in
Blockchain technology
German Ministry of Health is looking
for blockchain concepts too
9
© 2019 | IQVIA, All rights ReservedIl punto fondamentale è la reale capacità di creare valore, a fronte
del rischio di favorire semplicemente un’esplosione di dati
Format
Neural Networks
Remodeling Connect Real-Time Image
Monitoring Processing
Virtual Assistants
Document Decision Making
Tagging Pattern Recognition
Assignment Classificatio AI Doctor
n Augmented Reality
Smart Search
Data XML Virtual Reality Decision
Digital Cleansing Conversion Output Human Platform
Making
Interaction Integration
Transformation Quality Feedback Social, Web,
Digital
Engineering
Input
Smart NLP
Meta
Clinical, Medical
Formation
Workflow Meta-
Analysis
Natural Chatbots
Data RPA Language
Deconstruct Structure Tagging
Generation
Automated
Advanced Analytics Neural
Natural Reasoning
Disease Models Machine
Digital Lake Informatics Language Semantic
Real World Data Models Translation Ontologies
Population Health Models
Processing
Clinical Research
Investigator Protocols
Voice
Brochures
Informed Consent
Multiple Forms
languages Ethic Committee
Letters
Copyright © 2019 IQVIA. All rights reserved. IQVIATM is a registered trademark of IQVIA Inc. in the United States 10
and various other countries.Le best practice IQVIA indicano la strada degli “umbrella
frameworks”, capaci di indirizzare scelte clinico/sanitarie,
economico/operative e strategiche
INTEGRATED OPEN SECURE FUTURE-ORIENTED PHASED
• Administrative + • All clinical systems • Data locally in the • Knowledge Base • Big bang is not
Clinical • Room for own hospital • Support needed
• Hospital-wide development and • GDPR • Continuity in the • Data access is
• No silos external • Report-security product- gradually growing
• Structured data + applications • Application- development • Applications are
free text • BI agnostic security • Dialogue gradually growing
(Cognos, Qlik, …) • User group is
gradually growing
Combined analysis Flexible access Respect privacy/GDPR Reliability Scalability
11
© 2019 | IQVIA, All rights ReservedL’ambizione è interconnettere fonti diverse sfuttando piattaforme
capaci di gestire e sinconizzare ecosistemi di dati diversi
Electronic Medical Record | Clinical systems
Structured data
Unstructured data
Reports
Protocols
Notes
Discharge letters
...
Rapidly evolving context
Technologies
Applications
Potential partners
Government
…
12
© 2019 | IQVIA, All rights ReservedCASE STUDY
Il modello ospedaliero del futuro: un esempio pratico Hs PLATFORMS
INTEGRATED OPEN SECURE FUTURE-ORIENTED PHASED
REPORTS DASHBOARDS INDICATORs ANALYTICS BENCHMARKING SCHEDULING APPLICATION SERVER
PATIENT FINDER
APP
SEARCH ENGINE
Patiend Finder
Python Clinical Data Collector
text mining (web service)
DWH
Search- Export
ADMIN INTEGRATED DWH CLINICAL cohortes
engine
ELT BACKEND SERVER SEARCH SERVER
Gestructureerde data + Vrije text
SOURCE
ADT Invoicing ... LAB MEDICATIE ... Domain 1 Domain 2 ...
ADMINISTRATIVE SOURCE SYSTEMS EMR LEGACY CLINICAL SYSTEMS
13
© 2019 | IQVIA, All rights ReservedCASE STUDY
I benefici impattano tutti gli stakeholders del sistema Hs PLATFORMS
B E N E F I CI
PATIENT PHYSICIAN HOSPITAL PAYER INDUSTRY
• Higher quality of care • Easier access to • Faster access to • Improved visibility on • Stronger evidence in
(because of better his/her own EMR data enabling eligible patient reimbursement
access to outcome data efficiency gains population for drug dossiers
indicators) • View on outcome • Monitoring of treatment supporting • Better monitoring of
• Personalised and quality indicators (quality) indicators evidence based managed entry
medicine (e.g. PROMS - • Linked financial and reimbursement agreements
• Better treatment PREMS) clinical data for decision • Reduced costs in
decision • Faster patient financial optimization • Better view on R&D due to faster
• Ealier (rare) disease identification for • Faster response to resource use and patient identification
diagnosis clinical trials clinical trials cost by disease • More robust patient
• Faster clinical trial • More efficient set up improving chance to diagnosis and based brand
enrollment of patient cohorts be selected severity forecasts
• Better patient • Support in (rare) • Accelerate • Better monitoring of
experience disease diagnosis implementation of managed entry
the EMR agreements
14
© 2019 | IQVIA, All rights ReservedEsempio: applicazioni ML per l’identificazione CASE STUDY
di sotto-popolazioni per arruolamento in studi clinici RICERCA CLICNICA
MODELLO TRADIZIONALE
MODELLO «ENABLED» da ML
3
1
2
© 2019 | IQVIA, All rights Reserved
15CASE STUDY
I benefici calcolati in ambito R&D sono importanti RICERCA CLICNICA
143
105 13,8
75 20 7.1
Days Days Months
Faster Faster Faster
Months
Days
Days
6,7
123
30
Oncology Next
NextGen Industry
Industry Next
NextGen Traditional
Traditional Next
NextGen
Median generation
Method average
Average generation
Method method
Method generation
Method
approach approach approach
SITE IDENTIFICATION FIRST PATIENT ENROLLMENT
ENROLLED MONTHS
16
© 2019 | IQVIA, All rights ReservedModelli predittivi basati su cartelle cliniche ospedaliere per CASE STUDY
migliorare la diagnosi in malattie rare PRATICA CLINICA
Background Implementation and Impact
• Under-diagnosis is a major IQVIA built a data-driven predictive model to enable better detection of these hard-
issue and patients are often to-find patients and support earlier diagnosis and access to proper care
misdiagnosed
• Patients with this rare disease
have very high levels of activity
pre-diagnosis: on average, a
patient has 25 hospital events Capture complexities in data
in the 3 years prior to Data extract Examined the frequency, Identify key predictors
diagnosis. Used data from partnership timing, and number of Identified key primary and
with leading clinical centre and unique events defined by secondary diagnoses as
Challenge integrated with national clinical and activities’ codes to well as specialty visits that
• Above observations suggest hospital records to define predict patients earlier in differentiated patients with the
medical history data could be patients’ medical history based disease course disease versus those without
used to reveal undiagnosed on 700 prioritised clinical
codes
patients and detect patients
earlier in the disease course Deploy machine learning model
Performance of model exceeded KOL expectations –
discussions now focused on developing screening tool to be
deployed at specialty care clinical sites
© 2019 | IQVIA, All rights Reserved 17Applicazioni basate su analisi ed ottimizzazione del patient CASE STUDY
PP Ottimization
pathway, secondo un modello “Health Value Based”
REQUISITI ed OPPORTUNITA’
• Sistemi Informativi Digitali ed Interoperabili:
ospedale, territorio...ed oltre (Population Health Mngt)
• Regole chiare e trasparenti di accesso e trattamento
(GDPR compliant) per applicazioni Big Data cross-
platforms
• Apertura a sperimentare setting sanitari «orizzontali»
(dalle attività ai pazienti), ottimizzando ruoli e
REQUISITI competenze
• Puntare sull’eccellenza del sistema sanitario, come
fonte di attrazione risorse/investimenti Paese
• Consentire al privato di essere risorsa del sistema...
© 2019 | IQVIA, All rights Reserved
18CODE is the European multi-stakeholder,
collaborative group, led by IQVIA, that is supporting CASE STUDY
CODE
the creation of the Oncology Data Network (ODN)
Copyright © 2019 IQVIA, All Rights Reserved. Confidential and Proprietary. 19 19ILLUSTRATIVE – NOT TECHNICAL SCHEMATIC
CASE STUDY
We are building a flexible platform which extracts data from CODE
existing clinical systems, any European cancer treatment
centre and for all cancer types
THE ODN Hospital Level
Only an ODN Member can explore its
own patient data
Quality loop
Healthcare Analytics
Trusted Third Party (TPP)* Analytical outputs only n>=5
Non-identified data
Network Level
ODN Members can explore insights for different
patient groups from across the network
Sub-network A community to connect
European Data Warehouse* Country
Multi-country
Non-identified data Europe
Copyright © 2019 IQVIA, All Rights Reserved. Confidential and Proprietary. *Non-identified data is processed and stored in secure data centres which are certified by government agencies as required 9 20
in each country, or otherwise comply with industry securitystandardsCASE STUDY
The current footprint in Europe is growing rapidly with 110 CODE
member hospitals and first insights are being generated
110 11 Academic Hospitals
ODN Member hospitals*
99 Non-academic Hospitals
6+ 38 Public
Countries in scope 72 Private
77,000+
Est. Annual Patients onAnti-
ODN Members Cancer Therapy
Deployed ODN Members
*As of April 2019
Copyright © 2019 IQVIA, All Rights Reserved. Confidential and Proprietary. 21 21Luca Pinto Principal Real World Insights luca.pinto@iqvia.com +39 346 7224409
Puoi anche leggere
